3. aMODEL SOLUTIONS.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ec0795d2",
   "metadata": {},
   "source": [
    "# Import  Modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "af2fdb76",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\PC\\AppData\\Local\\Temp\\ipykernel_10344\\2334158970.py:6: DeprecationWarning: `import pandas_profiling` is going to be deprecated by April 1st. Please use `import ydata_profiling` instead.\n",
      "  import pandas_profiling\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "#pip install pandas-profiling\n",
    "import pandas_profiling\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "pd.set_option('display.max_columns', 100)\n",
    "pd.set_option('display.max_rows', 100)\n",
    "from sklearn.preprocessing import OrdinalEncoder,LabelEncoder,OrdinalEncoder,OneHotEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "from imblearn.over_sampling import SMOTE\n",
    "\n",
    "seed=0\n",
    "\n",
    "from sklearn.metrics import mean_squared_error,accuracy_score, roc_auc_score,f1_score,confusion_matrix\n",
    "from sklearn.preprocessing import OrdinalEncoder,LabelEncoder,OneHotEncoder\n",
    "from sklearn.preprocessing import StandardScaler, RobustScaler,MinMaxScaler,PolynomialFeatures\n",
    "from sklearn.model_selection import train_test_split, KFold\n",
    "\n",
    "from sklearn.ensemble import RandomForestRegressor,GradientBoostingRegressor\n",
    "from catboost import CatBoostRegressor,CatBoostClassifier\n",
    "from lightgbm import LGBMRegressor,LGBMClassifier\n",
    "from xgboost import XGBRegressor,XGBClassifier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "937c4413",
   "metadata": {},
   "source": [
    "# Read the Datasets"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1b5629a1",
   "metadata": {},
   "source": [
    "#### VALIDATION"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "70b83b2a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(175150, 11)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>B0017-25_24</td>\n",
       "      <td>2023-02-20 06:00:00</td>\n",
       "      <td>99.714558</td>\n",
       "      <td>8.678</td>\n",
       "      <td>3.715410</td>\n",
       "      <td>7.653169</td>\n",
       "      <td>5.964988</td>\n",
       "      <td>5.403557</td>\n",
       "      <td>6.815789</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>B0017-25_24</td>\n",
       "      <td>2023-02-20 07:00:00</td>\n",
       "      <td>99.927484</td>\n",
       "      <td>24.264</td>\n",
       "      <td>3.145402</td>\n",
       "      <td>10.015796</td>\n",
       "      <td>6.141206</td>\n",
       "      <td>6.544645</td>\n",
       "      <td>6.777114</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>B0017-25_24</td>\n",
       "      <td>2023-02-20 08:00:00</td>\n",
       "      <td>99.357688</td>\n",
       "      <td>84.004</td>\n",
       "      <td>2.923368</td>\n",
       "      <td>14.209819</td>\n",
       "      <td>5.582824</td>\n",
       "      <td>5.667775</td>\n",
       "      <td>1.175289</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>B0017-25_24</td>\n",
       "      <td>2023-02-20 09:00:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>301.0</td>\n",
       "      <td>0.654162</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>B0017-25_25</td>\n",
       "      <td>2023-02-20 19:00:00</td>\n",
       "      <td>99.264875</td>\n",
       "      <td>93.086</td>\n",
       "      <td>2.732594</td>\n",
       "      <td>16.947693</td>\n",
       "      <td>5.475508</td>\n",
       "      <td>4.719393</td>\n",
       "      <td>0.522540</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         NE ID             endTime  access_success_rate  \\\n",
       "0  B0017-25_24 2023-02-20 06:00:00            99.714558   \n",
       "1  B0017-25_24 2023-02-20 07:00:00            99.927484   \n",
       "2  B0017-25_24 2023-02-20 08:00:00            99.357688   \n",
       "3  B0017-25_24 2023-02-20 09:00:00                  NaN   \n",
       "4  B0017-25_25 2023-02-20 19:00:00            99.264875   \n",
       "\n",
       "   resource_utilition_rate        TA       bler       cqi       mcs  \\\n",
       "0                    8.678  3.715410   7.653169  5.964988  5.403557   \n",
       "1                   24.264  3.145402  10.015796  6.141206  6.544645   \n",
       "2                   84.004  2.923368  14.209819  5.582824  5.667775   \n",
       "3                      NaN       NaN        NaN       NaN       NaN   \n",
       "4                   93.086  2.732594  16.947693  5.475508  4.719393   \n",
       "\n",
       "   data_rate  fault_duration  relation  \n",
       "0   6.815789             0.0  0.000000  \n",
       "1   6.777114             0.0  0.000000  \n",
       "2   1.175289             0.0  0.000000  \n",
       "3        NaN           301.0  0.654162  \n",
       "4   0.522540             0.0  0.000000  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the testing data from the \"test.csv\" file and store it in the \"test\" DataFrame\n",
    "test = pd.read_csv(\"test.csv\")\n",
    "\n",
    "# Convert the \"endTime\" column in the \"test\" DataFrame to datetime format\n",
    "test['endTime'] = pd.to_datetime(test['endTime'])\n",
    "\n",
    "# Print the shape (number of rows and columns) of the \"test\" DataFrame\n",
    "print(test.shape)\n",
    "\n",
    "# Display the first few rows of the \"test\" DataFrame\n",
    "test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "284274f2",
   "metadata": {},
   "source": [
    "#### TRAIN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "5f5d8161",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(908922, 11)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>908002</th>\n",
       "      <td>B0729-13_1</td>\n",
       "      <td>2023-02-27 23:00:00</td>\n",
       "      <td>99.975315</td>\n",
       "      <td>44.245</td>\n",
       "      <td>3.244196</td>\n",
       "      <td>9.938242</td>\n",
       "      <td>9.106802</td>\n",
       "      <td>11.470092</td>\n",
       "      <td>19.380749</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>908003</th>\n",
       "      <td>B0729-13_1</td>\n",
       "      <td>2023-02-28 00:00:00</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>35.853</td>\n",
       "      <td>3.175133</td>\n",
       "      <td>10.320035</td>\n",
       "      <td>9.238970</td>\n",
       "      <td>11.740756</td>\n",
       "      <td>21.438768</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>908004</th>\n",
       "      <td>B0729-13_1</td>\n",
       "      <td>2023-02-28 01:00:00</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>31.238</td>\n",
       "      <td>3.196790</td>\n",
       "      <td>9.079722</td>\n",
       "      <td>9.034929</td>\n",
       "      <td>9.725438</td>\n",
       "      <td>22.493861</td>\n",
       "      <td>950.0</td>\n",
       "      <td>0.99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>908005</th>\n",
       "      <td>B0729-13_1</td>\n",
       "      <td>2023-02-28 02:00:00</td>\n",
       "      <td>99.547511</td>\n",
       "      <td>1.728</td>\n",
       "      <td>3.338323</td>\n",
       "      <td>6.922154</td>\n",
       "      <td>9.874873</td>\n",
       "      <td>8.773274</td>\n",
       "      <td>21.961071</td>\n",
       "      <td>209.0</td>\n",
       "      <td>0.99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>908006</th>\n",
       "      <td>B0729-15_1</td>\n",
       "      <td>2023-02-09 01:00:00</td>\n",
       "      <td>99.977350</td>\n",
       "      <td>29.442</td>\n",
       "      <td>3.055147</td>\n",
       "      <td>10.424411</td>\n",
       "      <td>7.210986</td>\n",
       "      <td>9.657070</td>\n",
       "      <td>8.758230</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             NE ID             endTime  access_success_rate  \\\n",
       "908002  B0729-13_1 2023-02-27 23:00:00            99.975315   \n",
       "908003  B0729-13_1 2023-02-28 00:00:00           100.000000   \n",
       "908004  B0729-13_1 2023-02-28 01:00:00           100.000000   \n",
       "908005  B0729-13_1 2023-02-28 02:00:00            99.547511   \n",
       "908006  B0729-15_1 2023-02-09 01:00:00            99.977350   \n",
       "\n",
       "        resource_utilition_rate        TA       bler       cqi        mcs  \\\n",
       "908002                   44.245  3.244196   9.938242  9.106802  11.470092   \n",
       "908003                   35.853  3.175133  10.320035  9.238970  11.740756   \n",
       "908004                   31.238  3.196790   9.079722  9.034929   9.725438   \n",
       "908005                    1.728  3.338323   6.922154  9.874873   8.773274   \n",
       "908006                   29.442  3.055147  10.424411  7.210986   9.657070   \n",
       "\n",
       "        data_rate  fault_duration  relation  \n",
       "908002  19.380749             0.0      0.00  \n",
       "908003  21.438768             0.0      0.00  \n",
       "908004  22.493861           950.0      0.99  \n",
       "908005  21.961071           209.0      0.99  \n",
       "908006   8.758230             0.0      0.00  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the training data from the \"train.csv\" file and store it in the \"train\" DataFrame\n",
    "train = pd.read_csv(\"train.csv\")\n",
    "\n",
    "# Convert the \"endTime\" column in the \"train\" DataFrame to datetime format\n",
    "train['endTime'] = pd.to_datetime(train['endTime'])\n",
    "\n",
    "# Print the shape (number of rows and columns) of the \"train\" DataFrame\n",
    "print(train.shape)\n",
    "\n",
    "# Retrieve the last 1000 rows from the \"train\" DataFrame and display rows 80 to 84 from that subset\n",
    "train.tail(1000)[80:85]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7534e24b",
   "metadata": {},
   "source": [
    "#### Combine Train and Test\n",
    "-For easier manipulation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "1a23885a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1084072\n"
     ]
    }
   ],
   "source": [
    "# Create a new column \"predict_rows\" in the train DataFrame and set all initial values to 0\n",
    "train[\"predict_rows\"] = 0\n",
    "\n",
    "# Create a new column \"predict_rows\" in the test DataFrame and set all initial values to 0\n",
    "test[\"predict_rows\"] = 0\n",
    "\n",
    "# Set the values in the \"predict_rows\" column of the test DataFrame to 1\n",
    "# for rows where the \"fault_duration\" column has a value greater than 0 for easy retrieval later\n",
    "test.loc[test[\"fault_duration\"] > 0, \"predict_rows\"] = 1\n",
    "\n",
    "# Concatenate the train and test DataFrames into a single DataFrame called \"combined_df\"\n",
    "combined_df = pd.concat([train, test]) \n",
    "\n",
    "# Sort the combined DataFrame by \"NE ID\" and \"endTime\" columns in ascending order\n",
    "combined_df.sort_values(by=[\"NE ID\", \"endTime\"], inplace=True)\n",
    "\n",
    "# Calculate and print the length of the combined DataFrame\n",
    "print(len(combined_df))\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "#combined_df[combined_df[\"NE ID\"]==\"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d038c038",
   "metadata": {},
   "source": [
    "# Create Target variable\n",
    "-If data rate of previous non fault row is greater than data rate of the fault row , then the label is 1 else 0 and that's our target."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "8ce5444a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>37561</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 05:00:00</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37562</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37563</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37564</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>99.829787</td>\n",
       "      <td>36.649</td>\n",
       "      <td>2.542375</td>\n",
       "      <td>12.062139</td>\n",
       "      <td>7.125136</td>\n",
       "      <td>7.467310</td>\n",
       "      <td>2.897868</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37565</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 09:00:00</td>\n",
       "      <td>99.860807</td>\n",
       "      <td>58.506</td>\n",
       "      <td>2.412703</td>\n",
       "      <td>12.039579</td>\n",
       "      <td>6.916188</td>\n",
       "      <td>7.704837</td>\n",
       "      <td>2.397028</td>\n",
       "      <td>57.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.172830</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37566</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 10:00:00</td>\n",
       "      <td>99.844781</td>\n",
       "      <td>50.280</td>\n",
       "      <td>2.428769</td>\n",
       "      <td>12.145683</td>\n",
       "      <td>7.321369</td>\n",
       "      <td>8.263317</td>\n",
       "      <td>3.663715</td>\n",
       "      <td>2938.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.528440</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37567</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 11:00:00</td>\n",
       "      <td>99.942639</td>\n",
       "      <td>46.411</td>\n",
       "      <td>2.510660</td>\n",
       "      <td>12.264153</td>\n",
       "      <td>7.599301</td>\n",
       "      <td>8.270461</td>\n",
       "      <td>5.753700</td>\n",
       "      <td>1089.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.570455</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37568</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 12:00:00</td>\n",
       "      <td>90.738391</td>\n",
       "      <td>60.956</td>\n",
       "      <td>2.461951</td>\n",
       "      <td>12.093170</td>\n",
       "      <td>7.612720</td>\n",
       "      <td>9.045568</td>\n",
       "      <td>6.245630</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.085498</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  access_success_rate  \\\n",
       "37561  B0017-13_8 2023-02-17 05:00:00           100.000000   \n",
       "37562  B0017-13_8 2023-02-17 06:00:00           100.000000   \n",
       "37563  B0017-13_8 2023-02-17 07:00:00            99.747899   \n",
       "37564  B0017-13_8 2023-02-17 08:00:00            99.829787   \n",
       "37565  B0017-13_8 2023-02-17 09:00:00            99.860807   \n",
       "37566  B0017-13_8 2023-02-17 10:00:00            99.844781   \n",
       "37567  B0017-13_8 2023-02-17 11:00:00            99.942639   \n",
       "37568  B0017-13_8 2023-02-17 12:00:00            90.738391   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "37561                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "37562                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "37563                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "37564                   36.649  2.542375  12.062139   7.125136  7.467310   \n",
       "37565                   58.506  2.412703  12.039579   6.916188  7.704837   \n",
       "37566                   50.280  2.428769  12.145683   7.321369  8.263317   \n",
       "37567                   46.411  2.510660  12.264153   7.599301  8.270461   \n",
       "37568                   60.956  2.461951  12.093170   7.612720  9.045568   \n",
       "\n",
       "       data_rate  fault_duration  relation  predict_rows  data_roc  \\\n",
       "37561  89.909177             0.0  0.000000             0       NaN   \n",
       "37562  48.475408             0.0  0.000000             0 -0.460840   \n",
       "37563   8.450350             0.0  0.000000             0 -0.825678   \n",
       "37564   2.897868            46.0  0.654162             0 -0.657071   \n",
       "37565   2.397028            57.0  0.654162             0 -0.172830   \n",
       "37566   3.663715          2938.0  0.654162             0  0.528440   \n",
       "37567   5.753700          1089.0  0.654162             0  0.570455   \n",
       "37568   6.245630            25.0  0.654162             0  0.085498   \n",
       "\n",
       "       data_rate_t+1_trend  \n",
       "37561                    1  \n",
       "37562                    1  \n",
       "37563                    1  \n",
       "37564                    1  \n",
       "37565                    1  \n",
       "37566                    0  \n",
       "37567                    0  \n",
       "37568                    0  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Calculate the percentage change in the 'data_rate' column for each group of 'NE ID'\n",
    "combined_df['data_roc'] = combined_df.groupby('NE ID', group_keys=False)['data_rate'].apply(lambda x: x.pct_change())\n",
    "\n",
    "# Define a function to determine the trend of data rate change\n",
    "def data_rate_change(group):\n",
    "    # Create a new column 'data_rate_t+1_trend' based on the percentage change\n",
    "    # If the percentage change is greater than 0, assign trend as 0 (no change or increase)\n",
    "    # Otherwise, assign trend as 1 (decrease)\n",
    "    group['data_rate_t+1_trend'] = group['data_roc'].apply(lambda x: 0 if x > 0 else 1)\n",
    "    return group\n",
    "\n",
    "# Group the data by 'NE ID', apply the 'data_rate_change' function, and maintain the structure\n",
    "train_grouped = combined_df.groupby('NE ID', as_index=False, group_keys=False).apply(data_rate_change)\n",
    "\n",
    "# Reset the index of the DataFrame after applying the transformation\n",
    "combined_df = train_grouped.reset_index(drop=True)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"]==\"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "deff6fec",
   "metadata": {},
   "source": [
    "# Data From Previous Hour to predict the data for the current hour\n",
    "-Achieved through shifting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "768d4fd2",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>37561</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 05:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37562</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37563</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37564</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37565</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 09:00:00</td>\n",
       "      <td>57.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.172830</td>\n",
       "      <td>1</td>\n",
       "      <td>99.829787</td>\n",
       "      <td>36.649</td>\n",
       "      <td>2.542375</td>\n",
       "      <td>12.062139</td>\n",
       "      <td>7.125136</td>\n",
       "      <td>7.467310</td>\n",
       "      <td>2.897868</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37566</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 10:00:00</td>\n",
       "      <td>2938.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.528440</td>\n",
       "      <td>0</td>\n",
       "      <td>99.860807</td>\n",
       "      <td>58.506</td>\n",
       "      <td>2.412703</td>\n",
       "      <td>12.039579</td>\n",
       "      <td>6.916188</td>\n",
       "      <td>7.704837</td>\n",
       "      <td>2.397028</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37567</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 11:00:00</td>\n",
       "      <td>1089.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.570455</td>\n",
       "      <td>0</td>\n",
       "      <td>99.844781</td>\n",
       "      <td>50.280</td>\n",
       "      <td>2.428769</td>\n",
       "      <td>12.145683</td>\n",
       "      <td>7.321369</td>\n",
       "      <td>8.263317</td>\n",
       "      <td>3.663715</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37568</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 12:00:00</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.085498</td>\n",
       "      <td>0</td>\n",
       "      <td>99.942639</td>\n",
       "      <td>46.411</td>\n",
       "      <td>2.510660</td>\n",
       "      <td>12.264153</td>\n",
       "      <td>7.599301</td>\n",
       "      <td>8.270461</td>\n",
       "      <td>5.753700</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "37561  B0017-13_8 2023-02-17 05:00:00             0.0  0.000000             0   \n",
       "37562  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "37563  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "37564  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "37565  B0017-13_8 2023-02-17 09:00:00            57.0  0.654162             0   \n",
       "37566  B0017-13_8 2023-02-17 10:00:00          2938.0  0.654162             0   \n",
       "37567  B0017-13_8 2023-02-17 11:00:00          1089.0  0.654162             0   \n",
       "37568  B0017-13_8 2023-02-17 12:00:00            25.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "37561       NaN                    1                  NaN   \n",
       "37562 -0.460840                    1           100.000000   \n",
       "37563 -0.825678                    1           100.000000   \n",
       "37564 -0.657071                    1            99.747899   \n",
       "37565 -0.172830                    1            99.829787   \n",
       "37566  0.528440                    0            99.860807   \n",
       "37567  0.570455                    0            99.844781   \n",
       "37568  0.085498                    0            99.942639   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "37561                      NaN       NaN        NaN        NaN       NaN   \n",
       "37562                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "37563                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "37564                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "37565                   36.649  2.542375  12.062139   7.125136  7.467310   \n",
       "37566                   58.506  2.412703  12.039579   6.916188  7.704837   \n",
       "37567                   50.280  2.428769  12.145683   7.321369  8.263317   \n",
       "37568                   46.411  2.510660  12.264153   7.599301  8.270461   \n",
       "\n",
       "       data_rate  \n",
       "37561        NaN  \n",
       "37562  89.909177  \n",
       "37563  48.475408  \n",
       "37564   8.450350  \n",
       "37565   2.897868  \n",
       "37566   2.397028  \n",
       "37567   3.663715  \n",
       "37568   5.753700  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a list of columns excluding those specified in the exclusion list\n",
    "cols = [col for col in train.columns if col not in [\"NE ID\", \"endTime\", \"fault_duration\", \"relation\",\\\n",
    "                                                    'data_rate_t+1_trend', \"predict_rows\"]]\n",
    "\n",
    "# Assign a unique group ID to each 'NE ID' group\n",
    "combined_df['Group_ID'] = combined_df.groupby('NE ID').ngroup() \n",
    "\n",
    "# Sort the combined DataFrame by 'Group_ID' and 'endTime'\n",
    "combined_df.sort_values(by=['Group_ID', 'endTime'], inplace=True) \n",
    "\n",
    "# Shift the specified columns within each 'Group_ID' to align with previous time steps\n",
    "shifted_data = combined_df.groupby(['Group_ID'], as_index=False)[cols].shift()\n",
    "\n",
    "# Drop the original columns and the 'Group_ID' column\n",
    "combined_df.drop(columns=cols+['Group_ID'], inplace=True)\n",
    "\n",
    "# Concatenate the original combined DataFrame with the shifted data\n",
    "combined_df = pd.concat([combined_df, shifted_data], axis=1)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "13454330",
   "metadata": {},
   "source": [
    "# Create Feature Fault"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "19dc4dd4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>37561</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 05:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37562</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37563</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37564</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37565</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 09:00:00</td>\n",
       "      <td>57.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.172830</td>\n",
       "      <td>1</td>\n",
       "      <td>99.829787</td>\n",
       "      <td>36.649</td>\n",
       "      <td>2.542375</td>\n",
       "      <td>12.062139</td>\n",
       "      <td>7.125136</td>\n",
       "      <td>7.467310</td>\n",
       "      <td>2.897868</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37566</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 10:00:00</td>\n",
       "      <td>2938.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.528440</td>\n",
       "      <td>0</td>\n",
       "      <td>99.860807</td>\n",
       "      <td>58.506</td>\n",
       "      <td>2.412703</td>\n",
       "      <td>12.039579</td>\n",
       "      <td>6.916188</td>\n",
       "      <td>7.704837</td>\n",
       "      <td>2.397028</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37567</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 11:00:00</td>\n",
       "      <td>1089.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.570455</td>\n",
       "      <td>0</td>\n",
       "      <td>99.844781</td>\n",
       "      <td>50.280</td>\n",
       "      <td>2.428769</td>\n",
       "      <td>12.145683</td>\n",
       "      <td>7.321369</td>\n",
       "      <td>8.263317</td>\n",
       "      <td>3.663715</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37568</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 12:00:00</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>0.085498</td>\n",
       "      <td>0</td>\n",
       "      <td>99.942639</td>\n",
       "      <td>46.411</td>\n",
       "      <td>2.510660</td>\n",
       "      <td>12.264153</td>\n",
       "      <td>7.599301</td>\n",
       "      <td>8.270461</td>\n",
       "      <td>5.753700</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "37561  B0017-13_8 2023-02-17 05:00:00             0.0  0.000000             0   \n",
       "37562  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "37563  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "37564  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "37565  B0017-13_8 2023-02-17 09:00:00            57.0  0.654162             0   \n",
       "37566  B0017-13_8 2023-02-17 10:00:00          2938.0  0.654162             0   \n",
       "37567  B0017-13_8 2023-02-17 11:00:00          1089.0  0.654162             0   \n",
       "37568  B0017-13_8 2023-02-17 12:00:00            25.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "37561       NaN                    1                  NaN   \n",
       "37562 -0.460840                    1           100.000000   \n",
       "37563 -0.825678                    1           100.000000   \n",
       "37564 -0.657071                    1            99.747899   \n",
       "37565 -0.172830                    1            99.829787   \n",
       "37566  0.528440                    0            99.860807   \n",
       "37567  0.570455                    0            99.844781   \n",
       "37568  0.085498                    0            99.942639   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "37561                      NaN       NaN        NaN        NaN       NaN   \n",
       "37562                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "37563                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "37564                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "37565                   36.649  2.542375  12.062139   7.125136  7.467310   \n",
       "37566                   58.506  2.412703  12.039579   6.916188  7.704837   \n",
       "37567                   50.280  2.428769  12.145683   7.321369  8.263317   \n",
       "37568                   46.411  2.510660  12.264153   7.599301  8.270461   \n",
       "\n",
       "       data_rate  fault  \n",
       "37561        NaN      0  \n",
       "37562  89.909177      0  \n",
       "37563  48.475408      0  \n",
       "37564   8.450350      1  \n",
       "37565   2.897868      1  \n",
       "37566   2.397028      1  \n",
       "37567   3.663715      1  \n",
       "37568   5.753700      1  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Define a function to create a fault label based on the \"fault_duration\" column\n",
    "def create_fault_label(row):\n",
    "    # If the \"fault_duration\" is greater than 0, label it as 1 (indicating fault)\n",
    "    # Otherwise, label it as 0 (indicating no fault)\n",
    "    if row[\"fault_duration\"] > 0:\n",
    "        return 1\n",
    "    else:\n",
    "        return 0\n",
    "\n",
    "# Apply the \"create_fault_label\" function to each row in the DataFrame using axis=1\n",
    "combined_df[\"fault\"] = combined_df.apply(create_fault_label, axis=1)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5dafe7f",
   "metadata": {},
   "source": [
    "# Reduce the Data "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6972f48f",
   "metadata": {},
   "source": [
    "#### ONLY PICK THE FIRST ROW OF DURATION>0\n",
    "-It was found that some rows had more than one instance of fault. For simplicity, I only pick the first instance of fault and ignore the rest."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "8d12a04c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: 1084072\n",
      "After: 995544\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>35624</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 05:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35625</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35626</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35627</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "35624  B0017-13_8 2023-02-17 05:00:00             0.0  0.000000             0   \n",
       "35625  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "35626  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "35627  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "35624       NaN                    1                  NaN   \n",
       "35625 -0.460840                    1           100.000000   \n",
       "35626 -0.825678                    1           100.000000   \n",
       "35627 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "35624                      NaN       NaN        NaN        NaN       NaN   \n",
       "35625                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "35626                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "35627                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  \n",
       "35624        NaN      0  \n",
       "35625  89.909177      0  \n",
       "35626  48.475408      0  \n",
       "35627   8.450350      1  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Print the length of the combined DataFrame before the filtering process\n",
    "before_filtering_length = len(combined_df)\n",
    "print(\"Before:\", before_filtering_length)\n",
    "\n",
    "# Group the DataFrame by 'NE ID'\n",
    "grouped = combined_df.groupby('NE ID')\n",
    "\n",
    "# Define a function to filter rows based on specific conditions\n",
    "def filter_rows(group):\n",
    "    # Select rows with 'fault_duration' equal to 0\n",
    "    rows_with_fault_duration_zero = group[group['fault_duration'] == 0]\n",
    "\n",
    "    # Find the index of the first row with 'fault_duration' greater than 0\n",
    "    first_fault_duration_gt_zero_index = group[group['fault_duration'] > 0].index[0]\n",
    "\n",
    "    # Select the row corresponding to the first fault duration greater than 0\n",
    "    row_with_first_fault_duration_gt_zero = group.loc[first_fault_duration_gt_zero_index:\\\n",
    "                                                      first_fault_duration_gt_zero_index]\n",
    "\n",
    "    # Concatenate the selected rows and return\n",
    "    return pd.concat([rows_with_fault_duration_zero, row_with_first_fault_duration_gt_zero])\n",
    "\n",
    "# Apply the filtering function to each group and reset the index\n",
    "combined_df = grouped.apply(filter_rows).reset_index(drop=True)\n",
    "\n",
    "# Print the length of the combined DataFrame after the filtering process\n",
    "after_filtering_length = len(combined_df)\n",
    "print(\"After:\", after_filtering_length)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f94734be",
   "metadata": {},
   "source": [
    "#### DROP ALL ROWS WITH 0 IN THE 'DATA_RATE' COLUMN AND 0 IN THE 'FAULT_DURATION' COLUMN\n",
    "-In typical scenarios, 'data_rate' would be 0 during faults; thus, deviations signal anomalies. After inspecting datataset, it was found that some had data_rate 0 and fault duration was also 0. Indicating an incosistency"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a5c04ef4",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: 995544\n",
      "After: 957480\n"
     ]
    }
   ],
   "source": [
    "# Print the length of the combined DataFrame before the filtering process\n",
    "before_filtering_length = len(combined_df)\n",
    "print(\"Before:\", before_filtering_length)\n",
    "\n",
    "# Filter the DataFrame based on conditions:\n",
    "# Select rows where 'data_rate' is not equal to 0 OR 'fault_duration' is not equal to 0. \n",
    "# In typical scenarios, 'data_rate' would be 0 during faults; thus, deviations signal anomalies.\n",
    "combined_df = combined_df.loc[(combined_df['data_rate'] != 0) | (combined_df['fault_duration'] != 0)]\n",
    "\n",
    "# Print the length of the combined DataFrame after the filtering process\n",
    "after_filtering_length = len(combined_df)\n",
    "print(\"After:\", after_filtering_length)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "#combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "76f72933",
   "metadata": {},
   "source": [
    "#### DROP THE ROWS AFFECTED BY SHIFTING\n",
    "-These rows are the first row of each ID and now contain NAN values after shifting the previous day."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "1fb20285",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before: 957480\n",
      "After: 948292\n"
     ]
    }
   ],
   "source": [
    "# Print the length of the combined DataFrame before the filtering process\n",
    "before_filtering_length = len(combined_df)\n",
    "print(\"Before:\", before_filtering_length)\n",
    "\n",
    "# Create a mask to identify rows where \"NE ID\" changes compared to the previous row\n",
    "mask = combined_df['NE ID'] != combined_df['NE ID'].shift()\n",
    "\n",
    "# Apply the mask to filter out rows where \"NE ID\" changes\n",
    "combined_df = combined_df[~mask]\n",
    "\n",
    "# Print the length of the combined DataFrame after the filtering process\n",
    "after_filtering_length = len(combined_df)\n",
    "print(\"After:\", after_filtering_length)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "32e1d0b8",
   "metadata": {},
   "source": [
    "# Create Time Features\n",
    "-The hour proved to be very important feature here!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "587d3ebb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>35625</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35626</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35627</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "35625  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "35626  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "35627  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "35625 -0.460840                    1           100.000000   \n",
       "35626 -0.825678                    1           100.000000   \n",
       "35627 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "35625                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "35626                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "35627                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \n",
       "35625  89.909177      0     6      2            4           1  \n",
       "35626  48.475408      0     7      2            4           1  \n",
       "35627   8.450350      1     8      2            4           1  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a new column \"hour\" in the combined DataFrame representing the hour of the day\n",
    "combined_df[\"hour\"] = combined_df[\"endTime\"].dt.hour\n",
    "\n",
    "# Create a new column \"month\" in the combined DataFrame representing the month of the year\n",
    "combined_df[\"month\"] = combined_df[\"endTime\"].dt.month\n",
    "\n",
    "# Create a new column \"day_of_week\" in the combined DataFrame representing the day of the week (0: Monday, 6: Sunday)\n",
    "combined_df[\"day_of_week\"] = combined_df[\"endTime\"].dt.dayofweek\n",
    "\n",
    "# Create a new column \"is_weekend\" in the combined DataFrame to indicate whether the day is a weekend (1) or not (0)\n",
    "combined_df[\"is_weekend\"] = combined_df[\"day_of_week\"].apply(lambda x: 1 if x < 5 else 0)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"]==\"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "f3f521da",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>hour_day_interaction</th>\n",
       "      <th>hour_month_interaction</th>\n",
       "      <th>hour_weekend_interaction</th>\n",
       "      <th>hour_day_month_interaction</th>\n",
       "      <th>period_of_day</th>\n",
       "      <th>hour_period_interaction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>35625</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>12</td>\n",
       "      <td>6</td>\n",
       "      <td>48</td>\n",
       "      <td>morning</td>\n",
       "      <td>24</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35626</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>56</td>\n",
       "      <td>morning</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35627</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>64</td>\n",
       "      <td>morning</td>\n",
       "      <td>32</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "35625  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "35626  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "35627  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "35625 -0.460840                    1           100.000000   \n",
       "35626 -0.825678                    1           100.000000   \n",
       "35627 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "35625                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "35626                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "35627                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \\\n",
       "35625  89.909177      0     6      2            4           1   \n",
       "35626  48.475408      0     7      2            4           1   \n",
       "35627   8.450350      1     8      2            4           1   \n",
       "\n",
       "       hour_day_interaction  hour_month_interaction  hour_weekend_interaction  \\\n",
       "35625                    24                      12                         6   \n",
       "35626                    28                      14                         7   \n",
       "35627                    32                      16                         8   \n",
       "\n",
       "       hour_day_month_interaction period_of_day  hour_period_interaction  \n",
       "35625                          48       morning                       24  \n",
       "35626                          56       morning                       28  \n",
       "35627                          64       morning                       32  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Create a new column \"hour_day_interaction\" by multiplying the 'hour' and 'day_of_week' columns\n",
    "combined_df['hour_day_interaction'] = combined_df['hour'] * combined_df['day_of_week']\n",
    "\n",
    "# Create a new column \"hour_month_interaction\" by multiplying the 'hour' and 'month' columns\n",
    "combined_df['hour_month_interaction'] = combined_df['hour'] * combined_df['month']\n",
    "\n",
    "# Create a new column \"hour_weekend_interaction\" by multiplying the 'hour' and 'is_weekend' columns\n",
    "combined_df['hour_weekend_interaction'] = combined_df['hour'] * combined_df['is_weekend']\n",
    "\n",
    "# Create a new column \"hour_day_month_interaction\" by multiplying the 'hour', 'day_of_week', and 'month' columns\n",
    "combined_df['hour_day_month_interaction'] = combined_df['hour'] * combined_df['day_of_week'] * combined_df['month']\n",
    "\n",
    "# Define a custom function to determine the period of the day based on the hour\n",
    "def get_period_of_day(hour):\n",
    "    if 0 <= hour < 6:\n",
    "        return 'night'\n",
    "    elif 6 <= hour < 12:\n",
    "        return 'morning'\n",
    "    elif 12 <= hour < 18:\n",
    "        return 'afternoon'\n",
    "    else:\n",
    "        return 'evening'\n",
    "\n",
    "# Apply the custom function to create the 'period_of_day' feature\n",
    "combined_df['period_of_day'] = combined_df['hour'].apply(get_period_of_day)\n",
    "\n",
    "#Create the interaction feature 'hour_period_interaction' by multiplying 'hour' with a weightin based on 'period_of_day'\n",
    "combined_df['hour_period_interaction'] = combined_df['hour'] * combined_df['period_of_day'].map({\n",
    "    'night': 1,\n",
    "    'morning': 4,\n",
    "    'afternoon': 3,\n",
    "    'evening': 2\n",
    "})\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "24ae10ae",
   "metadata": {},
   "source": [
    "# Length of Dataframe per ID Feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "44eff6b2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>hour_day_interaction</th>\n",
       "      <th>hour_month_interaction</th>\n",
       "      <th>hour_weekend_interaction</th>\n",
       "      <th>hour_day_month_interaction</th>\n",
       "      <th>period_of_day</th>\n",
       "      <th>hour_period_interaction</th>\n",
       "      <th>length_of_df</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>33858</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>12</td>\n",
       "      <td>6</td>\n",
       "      <td>48</td>\n",
       "      <td>morning</td>\n",
       "      <td>24</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33859</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>56</td>\n",
       "      <td>morning</td>\n",
       "      <td>28</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33860</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>64</td>\n",
       "      <td>morning</td>\n",
       "      <td>32</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "33858  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "33859  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "33860  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "33858 -0.460840                    1           100.000000   \n",
       "33859 -0.825678                    1           100.000000   \n",
       "33860 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "33858                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "33859                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "33860                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \\\n",
       "33858  89.909177      0     6      2            4           1   \n",
       "33859  48.475408      0     7      2            4           1   \n",
       "33860   8.450350      1     8      2            4           1   \n",
       "\n",
       "       hour_day_interaction  hour_month_interaction  hour_weekend_interaction  \\\n",
       "33858                    24                      12                         6   \n",
       "33859                    28                      14                         7   \n",
       "33860                    32                      16                         8   \n",
       "\n",
       "       hour_day_month_interaction period_of_day  hour_period_interaction  \\\n",
       "33858                          48       morning                       24   \n",
       "33859                          56       morning                       28   \n",
       "33860                          64       morning                       32   \n",
       "\n",
       "       length_of_df  \n",
       "33858             3  \n",
       "33859             3  \n",
       "33860             3  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Calculate the count of occurrences for each 'NE ID' in the combined DataFrame\n",
    "fault_duration_count = combined_df['NE ID'].value_counts().reset_index()\n",
    "\n",
    "# Rename the columns to 'NE ID' and 'length_of_df' for better clarity\n",
    "fault_duration_count.columns = ['NE ID', 'length_of_df']\n",
    "\n",
    "# Merge the calculated features with the original combined DataFrame based on 'NE ID'\n",
    "# This adds the 'length_of_df' column to the combined DataFrame\n",
    "combined_df = combined_df.merge(fault_duration_count, on='NE ID', how='left')\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9655f9a1",
   "metadata": {},
   "source": [
    "# Interaction Features of the KPI's"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "4127bde9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>hour_day_interaction</th>\n",
       "      <th>hour_month_interaction</th>\n",
       "      <th>hour_weekend_interaction</th>\n",
       "      <th>hour_day_month_interaction</th>\n",
       "      <th>period_of_day</th>\n",
       "      <th>hour_period_interaction</th>\n",
       "      <th>length_of_df</th>\n",
       "      <th>access_resource_ratio</th>\n",
       "      <th>bler_cqi_ratio</th>\n",
       "      <th>mcs_cqi_ratio</th>\n",
       "      <th>data_rate_access_difference</th>\n",
       "      <th>cqi_bler_difference</th>\n",
       "      <th>cqi_bler_product</th>\n",
       "      <th>access_mcs_product</th>\n",
       "      <th>access_mcs_ratio</th>\n",
       "      <th>bler_mcs_ratio</th>\n",
       "      <th>data_rate_resource_ratio</th>\n",
       "      <th>cqi_minus_bler</th>\n",
       "      <th>cqi_minus_mcs</th>\n",
       "      <th>access_rate_difference</th>\n",
       "      <th>resource_bler_product</th>\n",
       "      <th>data_rate_minus_mcs</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>33858</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>12</td>\n",
       "      <td>6</td>\n",
       "      <td>48</td>\n",
       "      <td>morning</td>\n",
       "      <td>24</td>\n",
       "      <td>3</td>\n",
       "      <td>94.876660</td>\n",
       "      <td>0.174597</td>\n",
       "      <td>0.269886</td>\n",
       "      <td>-10.090823</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>20.336202</td>\n",
       "      <td>291.270485</td>\n",
       "      <td>34.332349</td>\n",
       "      <td>0.646929</td>\n",
       "      <td>85.302824</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>7.879656</td>\n",
       "      <td>10.090823</td>\n",
       "      <td>1.986067</td>\n",
       "      <td>86.996472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33859</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>56</td>\n",
       "      <td>morning</td>\n",
       "      <td>28</td>\n",
       "      <td>3</td>\n",
       "      <td>58.754407</td>\n",
       "      <td>0.506945</td>\n",
       "      <td>0.841521</td>\n",
       "      <td>-51.524592</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>64.433691</td>\n",
       "      <td>948.727136</td>\n",
       "      <td>10.540438</td>\n",
       "      <td>0.602415</td>\n",
       "      <td>28.481438</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>1.786680</td>\n",
       "      <td>51.524592</td>\n",
       "      <td>9.727391</td>\n",
       "      <td>38.988137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33860</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>64</td>\n",
       "      <td>morning</td>\n",
       "      <td>32</td>\n",
       "      <td>3</td>\n",
       "      <td>6.869217</td>\n",
       "      <td>1.267927</td>\n",
       "      <td>1.090209</td>\n",
       "      <td>-91.297550</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>100.661536</td>\n",
       "      <td>968.942636</td>\n",
       "      <td>10.268558</td>\n",
       "      <td>1.163013</td>\n",
       "      <td>0.581940</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>-0.803771</td>\n",
       "      <td>91.297550</td>\n",
       "      <td>164.049673</td>\n",
       "      <td>-1.263566</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "33858  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "33859  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "33860  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "33858 -0.460840                    1           100.000000   \n",
       "33859 -0.825678                    1           100.000000   \n",
       "33860 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "33858                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "33859                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "33860                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \\\n",
       "33858  89.909177      0     6      2            4           1   \n",
       "33859  48.475408      0     7      2            4           1   \n",
       "33860   8.450350      1     8      2            4           1   \n",
       "\n",
       "       hour_day_interaction  hour_month_interaction  hour_weekend_interaction  \\\n",
       "33858                    24                      12                         6   \n",
       "33859                    28                      14                         7   \n",
       "33860                    32                      16                         8   \n",
       "\n",
       "       hour_day_month_interaction period_of_day  hour_period_interaction  \\\n",
       "33858                          48       morning                       24   \n",
       "33859                          56       morning                       28   \n",
       "33860                          64       morning                       32   \n",
       "\n",
       "       length_of_df  access_resource_ratio  bler_cqi_ratio  mcs_cqi_ratio  \\\n",
       "33858             3              94.876660        0.174597       0.269886   \n",
       "33859             3              58.754407        0.506945       0.841521   \n",
       "33860             3               6.869217        1.267927       1.090209   \n",
       "\n",
       "       data_rate_access_difference  cqi_bler_difference  cqi_bler_product  \\\n",
       "33858                   -10.090823             8.908046         20.336202   \n",
       "33859                   -51.524592             5.558681         64.433691   \n",
       "33860                   -91.297550            -2.387265        100.661536   \n",
       "\n",
       "       access_mcs_product  access_mcs_ratio  bler_mcs_ratio  \\\n",
       "33858          291.270485         34.332349        0.646929   \n",
       "33859          948.727136         10.540438        0.602415   \n",
       "33860          968.942636         10.268558        1.163013   \n",
       "\n",
       "       data_rate_resource_ratio  cqi_minus_bler  cqi_minus_mcs  \\\n",
       "33858                 85.302824        8.908046       7.879656   \n",
       "33859                 28.481438        5.558681       1.786680   \n",
       "33860                  0.581940       -2.387265      -0.803771   \n",
       "\n",
       "       access_rate_difference  resource_bler_product  data_rate_minus_mcs  \n",
       "33858               10.090823               1.986067            86.996472  \n",
       "33859               51.524592               9.727391            38.988137  \n",
       "33860               91.297550             164.049673            -1.263566  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Calculate ratios between KPIs\n",
    "combined_df['access_resource_ratio'] = combined_df['access_success_rate'] /(combined_df['resource_utilition_rate']+1e-9)\n",
    "combined_df['bler_cqi_ratio'] = combined_df['bler'] / (combined_df['cqi']+1e-9)\n",
    "combined_df['mcs_cqi_ratio'] = combined_df['mcs'] / (combined_df['cqi']+1e-9)\n",
    "\n",
    "# Calculate differences between KPIs\n",
    "combined_df['data_rate_access_difference'] = combined_df['data_rate'] - combined_df['access_success_rate']\n",
    "combined_df['cqi_bler_difference'] = combined_df['cqi'] - combined_df['bler']\n",
    "\n",
    "# Calculate products of KPIs\n",
    "combined_df['cqi_bler_product'] = combined_df['cqi'] * combined_df['bler']\n",
    "combined_df['access_mcs_product'] = combined_df['access_success_rate'] * combined_df['mcs']\n",
    "\n",
    "# Calculate more features based on existing KPIs\n",
    "combined_df['access_mcs_ratio'] = combined_df['access_success_rate'] / (combined_df['mcs'] + 1e-9)\n",
    "combined_df['bler_mcs_ratio'] = combined_df['bler'] / (combined_df['mcs'] + 1e-9)\n",
    "combined_df['data_rate_resource_ratio'] = combined_df['data_rate'] / (combined_df['resource_utilition_rate'] + 1e-9)\n",
    "\n",
    "combined_df['cqi_minus_bler'] = combined_df['cqi'] - combined_df['bler']\n",
    "combined_df['cqi_minus_mcs'] = combined_df['cqi'] - combined_df['mcs']\n",
    "combined_df['access_rate_difference'] = combined_df['access_success_rate'] - combined_df['data_rate']\n",
    "combined_df['resource_bler_product'] = combined_df['resource_utilition_rate'] * combined_df['bler']\n",
    "combined_df['data_rate_minus_mcs'] = combined_df['data_rate'] - combined_df['mcs']\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a8708682",
   "metadata": {},
   "source": [
    "# Bin Features\n",
    "#### Create Bins for Column \"relation\" and \"fault_duration\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "fa41087a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>hour_day_interaction</th>\n",
       "      <th>hour_month_interaction</th>\n",
       "      <th>hour_weekend_interaction</th>\n",
       "      <th>hour_day_month_interaction</th>\n",
       "      <th>period_of_day</th>\n",
       "      <th>hour_period_interaction</th>\n",
       "      <th>length_of_df</th>\n",
       "      <th>access_resource_ratio</th>\n",
       "      <th>bler_cqi_ratio</th>\n",
       "      <th>mcs_cqi_ratio</th>\n",
       "      <th>data_rate_access_difference</th>\n",
       "      <th>cqi_bler_difference</th>\n",
       "      <th>cqi_bler_product</th>\n",
       "      <th>access_mcs_product</th>\n",
       "      <th>access_mcs_ratio</th>\n",
       "      <th>bler_mcs_ratio</th>\n",
       "      <th>data_rate_resource_ratio</th>\n",
       "      <th>cqi_minus_bler</th>\n",
       "      <th>cqi_minus_mcs</th>\n",
       "      <th>access_rate_difference</th>\n",
       "      <th>resource_bler_product</th>\n",
       "      <th>data_rate_minus_mcs</th>\n",
       "      <th>relation_bin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>33858</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>12</td>\n",
       "      <td>6</td>\n",
       "      <td>48</td>\n",
       "      <td>morning</td>\n",
       "      <td>24</td>\n",
       "      <td>3</td>\n",
       "      <td>94.876660</td>\n",
       "      <td>0.174597</td>\n",
       "      <td>0.269886</td>\n",
       "      <td>-10.090823</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>20.336202</td>\n",
       "      <td>291.270485</td>\n",
       "      <td>34.332349</td>\n",
       "      <td>0.646929</td>\n",
       "      <td>85.302824</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>7.879656</td>\n",
       "      <td>10.090823</td>\n",
       "      <td>1.986067</td>\n",
       "      <td>86.996472</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33859</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>56</td>\n",
       "      <td>morning</td>\n",
       "      <td>28</td>\n",
       "      <td>3</td>\n",
       "      <td>58.754407</td>\n",
       "      <td>0.506945</td>\n",
       "      <td>0.841521</td>\n",
       "      <td>-51.524592</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>64.433691</td>\n",
       "      <td>948.727136</td>\n",
       "      <td>10.540438</td>\n",
       "      <td>0.602415</td>\n",
       "      <td>28.481438</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>1.786680</td>\n",
       "      <td>51.524592</td>\n",
       "      <td>9.727391</td>\n",
       "      <td>38.988137</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33860</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>64</td>\n",
       "      <td>morning</td>\n",
       "      <td>32</td>\n",
       "      <td>3</td>\n",
       "      <td>6.869217</td>\n",
       "      <td>1.267927</td>\n",
       "      <td>1.090209</td>\n",
       "      <td>-91.297550</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>100.661536</td>\n",
       "      <td>968.942636</td>\n",
       "      <td>10.268558</td>\n",
       "      <td>1.163013</td>\n",
       "      <td>0.581940</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>-0.803771</td>\n",
       "      <td>91.297550</td>\n",
       "      <td>164.049673</td>\n",
       "      <td>-1.263566</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "33858  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "33859  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "33860  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "33858 -0.460840                    1           100.000000   \n",
       "33859 -0.825678                    1           100.000000   \n",
       "33860 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "33858                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "33859                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "33860                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \\\n",
       "33858  89.909177      0     6      2            4           1   \n",
       "33859  48.475408      0     7      2            4           1   \n",
       "33860   8.450350      1     8      2            4           1   \n",
       "\n",
       "       hour_day_interaction  hour_month_interaction  hour_weekend_interaction  \\\n",
       "33858                    24                      12                         6   \n",
       "33859                    28                      14                         7   \n",
       "33860                    32                      16                         8   \n",
       "\n",
       "       hour_day_month_interaction period_of_day  hour_period_interaction  \\\n",
       "33858                          48       morning                       24   \n",
       "33859                          56       morning                       28   \n",
       "33860                          64       morning                       32   \n",
       "\n",
       "       length_of_df  access_resource_ratio  bler_cqi_ratio  mcs_cqi_ratio  \\\n",
       "33858             3              94.876660        0.174597       0.269886   \n",
       "33859             3              58.754407        0.506945       0.841521   \n",
       "33860             3               6.869217        1.267927       1.090209   \n",
       "\n",
       "       data_rate_access_difference  cqi_bler_difference  cqi_bler_product  \\\n",
       "33858                   -10.090823             8.908046         20.336202   \n",
       "33859                   -51.524592             5.558681         64.433691   \n",
       "33860                   -91.297550            -2.387265        100.661536   \n",
       "\n",
       "       access_mcs_product  access_mcs_ratio  bler_mcs_ratio  \\\n",
       "33858          291.270485         34.332349        0.646929   \n",
       "33859          948.727136         10.540438        0.602415   \n",
       "33860          968.942636         10.268558        1.163013   \n",
       "\n",
       "       data_rate_resource_ratio  cqi_minus_bler  cqi_minus_mcs  \\\n",
       "33858                 85.302824        8.908046       7.879656   \n",
       "33859                 28.481438        5.558681       1.786680   \n",
       "33860                  0.581940       -2.387265      -0.803771   \n",
       "\n",
       "       access_rate_difference  resource_bler_product  data_rate_minus_mcs  \\\n",
       "33858               10.090823               1.986067            86.996472   \n",
       "33859               51.524592               9.727391            38.988137   \n",
       "33860               91.297550             164.049673            -1.263566   \n",
       "\n",
       "       relation_bin  \n",
       "33858            -1  \n",
       "33859            -1  \n",
       "33860             2  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Define bins for the 'relation' values\n",
    "bins = [0, 0.25, 0.5, 0.75, 1]\n",
    "\n",
    "# Create a new column 'relation_bin' to categorize 'relation' values into bins\n",
    "combined_df['relation_bin'] = pd.cut(combined_df['relation'], bins=bins, labels=[0, 1, 2, 3])\n",
    "\n",
    "# Add a new category (-1) to the 'relation_bin' column for potential NaN values\n",
    "combined_df['relation_bin'] = combined_df['relation_bin'].cat.add_categories(-1)\n",
    "\n",
    "# Fill NaN values in the 'relation_bin' column with the new category (-1)\n",
    "combined_df['relation_bin'] = combined_df['relation_bin'].fillna(-1)\n",
    "\n",
    "# Convert the 'relation_bin' column to integer type\n",
    "combined_df['relation_bin'] = combined_df['relation_bin'].astype(int)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "7f89ccbb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>hour_day_interaction</th>\n",
       "      <th>hour_month_interaction</th>\n",
       "      <th>hour_weekend_interaction</th>\n",
       "      <th>hour_day_month_interaction</th>\n",
       "      <th>period_of_day</th>\n",
       "      <th>hour_period_interaction</th>\n",
       "      <th>length_of_df</th>\n",
       "      <th>access_resource_ratio</th>\n",
       "      <th>bler_cqi_ratio</th>\n",
       "      <th>mcs_cqi_ratio</th>\n",
       "      <th>data_rate_access_difference</th>\n",
       "      <th>cqi_bler_difference</th>\n",
       "      <th>cqi_bler_product</th>\n",
       "      <th>access_mcs_product</th>\n",
       "      <th>access_mcs_ratio</th>\n",
       "      <th>bler_mcs_ratio</th>\n",
       "      <th>data_rate_resource_ratio</th>\n",
       "      <th>cqi_minus_bler</th>\n",
       "      <th>cqi_minus_mcs</th>\n",
       "      <th>access_rate_difference</th>\n",
       "      <th>resource_bler_product</th>\n",
       "      <th>data_rate_minus_mcs</th>\n",
       "      <th>relation_bin</th>\n",
       "      <th>fault_duration_bin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>33858</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>12</td>\n",
       "      <td>6</td>\n",
       "      <td>48</td>\n",
       "      <td>morning</td>\n",
       "      <td>24</td>\n",
       "      <td>3</td>\n",
       "      <td>94.876660</td>\n",
       "      <td>0.174597</td>\n",
       "      <td>0.269886</td>\n",
       "      <td>-10.090823</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>20.336202</td>\n",
       "      <td>291.270485</td>\n",
       "      <td>34.332349</td>\n",
       "      <td>0.646929</td>\n",
       "      <td>85.302824</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>7.879656</td>\n",
       "      <td>10.090823</td>\n",
       "      <td>1.986067</td>\n",
       "      <td>86.996472</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33859</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>56</td>\n",
       "      <td>morning</td>\n",
       "      <td>28</td>\n",
       "      <td>3</td>\n",
       "      <td>58.754407</td>\n",
       "      <td>0.506945</td>\n",
       "      <td>0.841521</td>\n",
       "      <td>-51.524592</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>64.433691</td>\n",
       "      <td>948.727136</td>\n",
       "      <td>10.540438</td>\n",
       "      <td>0.602415</td>\n",
       "      <td>28.481438</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>1.786680</td>\n",
       "      <td>51.524592</td>\n",
       "      <td>9.727391</td>\n",
       "      <td>38.988137</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33860</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>64</td>\n",
       "      <td>morning</td>\n",
       "      <td>32</td>\n",
       "      <td>3</td>\n",
       "      <td>6.869217</td>\n",
       "      <td>1.267927</td>\n",
       "      <td>1.090209</td>\n",
       "      <td>-91.297550</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>100.661536</td>\n",
       "      <td>968.942636</td>\n",
       "      <td>10.268558</td>\n",
       "      <td>1.163013</td>\n",
       "      <td>0.581940</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>-0.803771</td>\n",
       "      <td>91.297550</td>\n",
       "      <td>164.049673</td>\n",
       "      <td>-1.263566</td>\n",
       "      <td>2</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "33858  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "33859  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "33860  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "33858 -0.460840                    1           100.000000   \n",
       "33859 -0.825678                    1           100.000000   \n",
       "33860 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "33858                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "33859                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "33860                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \\\n",
       "33858  89.909177      0     6      2            4           1   \n",
       "33859  48.475408      0     7      2            4           1   \n",
       "33860   8.450350      1     8      2            4           1   \n",
       "\n",
       "       hour_day_interaction  hour_month_interaction  hour_weekend_interaction  \\\n",
       "33858                    24                      12                         6   \n",
       "33859                    28                      14                         7   \n",
       "33860                    32                      16                         8   \n",
       "\n",
       "       hour_day_month_interaction period_of_day  hour_period_interaction  \\\n",
       "33858                          48       morning                       24   \n",
       "33859                          56       morning                       28   \n",
       "33860                          64       morning                       32   \n",
       "\n",
       "       length_of_df  access_resource_ratio  bler_cqi_ratio  mcs_cqi_ratio  \\\n",
       "33858             3              94.876660        0.174597       0.269886   \n",
       "33859             3              58.754407        0.506945       0.841521   \n",
       "33860             3               6.869217        1.267927       1.090209   \n",
       "\n",
       "       data_rate_access_difference  cqi_bler_difference  cqi_bler_product  \\\n",
       "33858                   -10.090823             8.908046         20.336202   \n",
       "33859                   -51.524592             5.558681         64.433691   \n",
       "33860                   -91.297550            -2.387265        100.661536   \n",
       "\n",
       "       access_mcs_product  access_mcs_ratio  bler_mcs_ratio  \\\n",
       "33858          291.270485         34.332349        0.646929   \n",
       "33859          948.727136         10.540438        0.602415   \n",
       "33860          968.942636         10.268558        1.163013   \n",
       "\n",
       "       data_rate_resource_ratio  cqi_minus_bler  cqi_minus_mcs  \\\n",
       "33858                 85.302824        8.908046       7.879656   \n",
       "33859                 28.481438        5.558681       1.786680   \n",
       "33860                  0.581940       -2.387265      -0.803771   \n",
       "\n",
       "       access_rate_difference  resource_bler_product  data_rate_minus_mcs  \\\n",
       "33858               10.090823               1.986067            86.996472   \n",
       "33859               51.524592               9.727391            38.988137   \n",
       "33860               91.297550             164.049673            -1.263566   \n",
       "\n",
       "       relation_bin  fault_duration_bin  \n",
       "33858            -1                  -1  \n",
       "33859            -1                  -1  \n",
       "33860             2                  -1  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Define bins for the 'fault_duration' values\n",
    "bins = [900, 1800, 2700, 3000]\n",
    "\n",
    "# Create a new column 'fault_duration_bin' to categorize 'fault_duration' values into bins\n",
    "combined_df['fault_duration_bin'] = pd.cut(combined_df['fault_duration'], bins=bins, labels=[0, 1, 2])\n",
    "\n",
    "# Add a new category (-1) to the 'fault_duration_bin' column for potential NaN values\n",
    "combined_df['fault_duration_bin'] = combined_df['fault_duration_bin'].cat.add_categories(-1)\n",
    "\n",
    "# Fill NaN values in the 'fault_duration_bin' column with the new category (-1)\n",
    "combined_df['fault_duration_bin'] = combined_df['fault_duration_bin'].fillna(-1)\n",
    "\n",
    "# Convert the 'fault_duration_bin' column to integer type\n",
    "combined_df['fault_duration_bin'] = combined_df['fault_duration_bin'].astype(int)\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "366fd266",
   "metadata": {},
   "source": [
    "#### Create Bins for \"Data_Roc\" "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "2dd03386",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\PC\\anaconda3\\lib\\site-packages\\numpy\\lib\\function_base.py:4527: RuntimeWarning: invalid value encountered in subtract\n",
      "  diff_b_a = subtract(b, a)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NE ID</th>\n",
       "      <th>endTime</th>\n",
       "      <th>fault_duration</th>\n",
       "      <th>relation</th>\n",
       "      <th>predict_rows</th>\n",
       "      <th>data_roc</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "      <th>access_success_rate</th>\n",
       "      <th>resource_utilition_rate</th>\n",
       "      <th>TA</th>\n",
       "      <th>bler</th>\n",
       "      <th>cqi</th>\n",
       "      <th>mcs</th>\n",
       "      <th>data_rate</th>\n",
       "      <th>fault</th>\n",
       "      <th>hour</th>\n",
       "      <th>month</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>is_weekend</th>\n",
       "      <th>hour_day_interaction</th>\n",
       "      <th>hour_month_interaction</th>\n",
       "      <th>hour_weekend_interaction</th>\n",
       "      <th>hour_day_month_interaction</th>\n",
       "      <th>period_of_day</th>\n",
       "      <th>hour_period_interaction</th>\n",
       "      <th>length_of_df</th>\n",
       "      <th>access_resource_ratio</th>\n",
       "      <th>bler_cqi_ratio</th>\n",
       "      <th>mcs_cqi_ratio</th>\n",
       "      <th>data_rate_access_difference</th>\n",
       "      <th>cqi_bler_difference</th>\n",
       "      <th>cqi_bler_product</th>\n",
       "      <th>access_mcs_product</th>\n",
       "      <th>access_mcs_ratio</th>\n",
       "      <th>bler_mcs_ratio</th>\n",
       "      <th>data_rate_resource_ratio</th>\n",
       "      <th>cqi_minus_bler</th>\n",
       "      <th>cqi_minus_mcs</th>\n",
       "      <th>access_rate_difference</th>\n",
       "      <th>resource_bler_product</th>\n",
       "      <th>data_rate_minus_mcs</th>\n",
       "      <th>relation_bin</th>\n",
       "      <th>fault_duration_bin</th>\n",
       "      <th>data_roc_bin</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>33858</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 06:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.460840</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.054</td>\n",
       "      <td>3.035211</td>\n",
       "      <td>1.884315</td>\n",
       "      <td>10.792361</td>\n",
       "      <td>2.912705</td>\n",
       "      <td>89.909177</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>12</td>\n",
       "      <td>6</td>\n",
       "      <td>48</td>\n",
       "      <td>morning</td>\n",
       "      <td>24</td>\n",
       "      <td>3</td>\n",
       "      <td>94.876660</td>\n",
       "      <td>0.174597</td>\n",
       "      <td>0.269886</td>\n",
       "      <td>-10.090823</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>20.336202</td>\n",
       "      <td>291.270485</td>\n",
       "      <td>34.332349</td>\n",
       "      <td>0.646929</td>\n",
       "      <td>85.302824</td>\n",
       "      <td>8.908046</td>\n",
       "      <td>7.879656</td>\n",
       "      <td>10.090823</td>\n",
       "      <td>1.986067</td>\n",
       "      <td>86.996472</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33859</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 07:00:00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.825678</td>\n",
       "      <td>1</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>1.702</td>\n",
       "      <td>2.638554</td>\n",
       "      <td>5.715271</td>\n",
       "      <td>11.273952</td>\n",
       "      <td>9.487271</td>\n",
       "      <td>48.475408</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>28</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>56</td>\n",
       "      <td>morning</td>\n",
       "      <td>28</td>\n",
       "      <td>3</td>\n",
       "      <td>58.754407</td>\n",
       "      <td>0.506945</td>\n",
       "      <td>0.841521</td>\n",
       "      <td>-51.524592</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>64.433691</td>\n",
       "      <td>948.727136</td>\n",
       "      <td>10.540438</td>\n",
       "      <td>0.602415</td>\n",
       "      <td>28.481438</td>\n",
       "      <td>5.558681</td>\n",
       "      <td>1.786680</td>\n",
       "      <td>51.524592</td>\n",
       "      <td>9.727391</td>\n",
       "      <td>38.988137</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33860</th>\n",
       "      <td>B0017-13_8</td>\n",
       "      <td>2023-02-17 08:00:00</td>\n",
       "      <td>46.0</td>\n",
       "      <td>0.654162</td>\n",
       "      <td>0</td>\n",
       "      <td>-0.657071</td>\n",
       "      <td>1</td>\n",
       "      <td>99.747899</td>\n",
       "      <td>14.521</td>\n",
       "      <td>2.963437</td>\n",
       "      <td>11.297409</td>\n",
       "      <td>8.910144</td>\n",
       "      <td>9.713915</td>\n",
       "      <td>8.450350</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>64</td>\n",
       "      <td>morning</td>\n",
       "      <td>32</td>\n",
       "      <td>3</td>\n",
       "      <td>6.869217</td>\n",
       "      <td>1.267927</td>\n",
       "      <td>1.090209</td>\n",
       "      <td>-91.297550</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>100.661536</td>\n",
       "      <td>968.942636</td>\n",
       "      <td>10.268558</td>\n",
       "      <td>1.163013</td>\n",
       "      <td>0.581940</td>\n",
       "      <td>-2.387265</td>\n",
       "      <td>-0.803771</td>\n",
       "      <td>91.297550</td>\n",
       "      <td>164.049673</td>\n",
       "      <td>-1.263566</td>\n",
       "      <td>2</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            NE ID             endTime  fault_duration  relation  predict_rows  \\\n",
       "33858  B0017-13_8 2023-02-17 06:00:00             0.0  0.000000             0   \n",
       "33859  B0017-13_8 2023-02-17 07:00:00             0.0  0.000000             0   \n",
       "33860  B0017-13_8 2023-02-17 08:00:00            46.0  0.654162             0   \n",
       "\n",
       "       data_roc  data_rate_t+1_trend  access_success_rate  \\\n",
       "33858 -0.460840                    1           100.000000   \n",
       "33859 -0.825678                    1           100.000000   \n",
       "33860 -0.657071                    1            99.747899   \n",
       "\n",
       "       resource_utilition_rate        TA       bler        cqi       mcs  \\\n",
       "33858                    1.054  3.035211   1.884315  10.792361  2.912705   \n",
       "33859                    1.702  2.638554   5.715271  11.273952  9.487271   \n",
       "33860                   14.521  2.963437  11.297409   8.910144  9.713915   \n",
       "\n",
       "       data_rate  fault  hour  month  day_of_week  is_weekend  \\\n",
       "33858  89.909177      0     6      2            4           1   \n",
       "33859  48.475408      0     7      2            4           1   \n",
       "33860   8.450350      1     8      2            4           1   \n",
       "\n",
       "       hour_day_interaction  hour_month_interaction  hour_weekend_interaction  \\\n",
       "33858                    24                      12                         6   \n",
       "33859                    28                      14                         7   \n",
       "33860                    32                      16                         8   \n",
       "\n",
       "       hour_day_month_interaction period_of_day  hour_period_interaction  \\\n",
       "33858                          48       morning                       24   \n",
       "33859                          56       morning                       28   \n",
       "33860                          64       morning                       32   \n",
       "\n",
       "       length_of_df  access_resource_ratio  bler_cqi_ratio  mcs_cqi_ratio  \\\n",
       "33858             3              94.876660        0.174597       0.269886   \n",
       "33859             3              58.754407        0.506945       0.841521   \n",
       "33860             3               6.869217        1.267927       1.090209   \n",
       "\n",
       "       data_rate_access_difference  cqi_bler_difference  cqi_bler_product  \\\n",
       "33858                   -10.090823             8.908046         20.336202   \n",
       "33859                   -51.524592             5.558681         64.433691   \n",
       "33860                   -91.297550            -2.387265        100.661536   \n",
       "\n",
       "       access_mcs_product  access_mcs_ratio  bler_mcs_ratio  \\\n",
       "33858          291.270485         34.332349        0.646929   \n",
       "33859          948.727136         10.540438        0.602415   \n",
       "33860          968.942636         10.268558        1.163013   \n",
       "\n",
       "       data_rate_resource_ratio  cqi_minus_bler  cqi_minus_mcs  \\\n",
       "33858                 85.302824        8.908046       7.879656   \n",
       "33859                 28.481438        5.558681       1.786680   \n",
       "33860                  0.581940       -2.387265      -0.803771   \n",
       "\n",
       "       access_rate_difference  resource_bler_product  data_rate_minus_mcs  \\\n",
       "33858               10.090823               1.986067            86.996472   \n",
       "33859               51.524592               9.727391            38.988137   \n",
       "33860               91.297550             164.049673            -1.263566   \n",
       "\n",
       "       relation_bin  fault_duration_bin  data_roc_bin  \n",
       "33858            -1                  -1             0  \n",
       "33859            -1                  -1             0  \n",
       "33860             2                  -1             0  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Define the number of bins for quantile binning\n",
    "num_bins = 3\n",
    "\n",
    "# Fill NaN values in the 'data_roc' column with a placeholder value (-999)\n",
    "combined_df['data_roc'].fillna(-999, inplace=True)\n",
    "\n",
    "# Perform quantile binning on the 'data_roc' column to create 'data_roc_bin' column\n",
    "quantile_breaks_combined_df = pd.qcut(combined_df['data_roc'], q=num_bins, labels=False, duplicates='drop')\n",
    "\n",
    "# Replace the value representing the number of bins with NaN\n",
    "quantile_breaks_combined_df.replace(num_bins, np.nan, inplace=True)\n",
    "\n",
    "# Assign the quantile bin labels to the 'data_roc_bin' column\n",
    "combined_df['data_roc_bin'] = quantile_breaks_combined_df\n",
    "\n",
    "# Display the rows in the combined DataFrame where the \"NE ID\" is \"B0017-13_8\"\n",
    "combined_df[combined_df[\"NE ID\"] == \"B0017-13_8\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "7127a561",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1    3789\n",
      "0    3435\n",
      "Name: data_rate_t+1_trend, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Display the combined DataFrame for rows where \"fault\" is 1 and \"predict_rows\" is 0, to check distribution of 1s and 0s\n",
    "# and then count the occurrences of each value in the \"data_rate_t+1_trend\" column\n",
    "trend_counts = combined_df[((combined_df[\"fault\"] == 1) & \\\n",
    "                            (combined_df[\"predict_rows\"] == 0))][\"data_rate_t+1_trend\"].value_counts()\n",
    "print(trend_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "22b7a185",
   "metadata": {},
   "source": [
    "# Modelling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "84cc8723",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\PC\\AppData\\Local\\Temp\\ipykernel_10344\\1718870088.py:19: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  train2.drop_duplicates(inplace=True)\n",
      "C:\\Users\\PC\\AppData\\Local\\Temp\\ipykernel_10344\\1718870088.py:23: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  train2.drop(columns=[\"data_rate_t+1_trend\", \"NE ID\", \"endTime\"], inplace=True, axis=1)\n",
      "C:\\Users\\PC\\AppData\\Local\\Temp\\ipykernel_10344\\1718870088.py:24: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  train2.drop(columns=drop_cols, inplace=True, axis=1)\n"
     ]
    }
   ],
   "source": [
    "# Create a copy of the combined DataFrame and assign it to the \"train\" variable\n",
    "train = combined_df.copy()\n",
    "\n",
    "# Create a \"test\" DataFrame containing rows where \"predict_rows\" is equal to 1\n",
    "test = train[train[\"predict_rows\"] == 1]\n",
    "\n",
    "# Create a \"train2\" DataFrame containing rows where \"predict_rows\" is equal to 0\n",
    "train2 = train[train[\"predict_rows\"] == 0]\n",
    "\n",
    "# List of columns to drop from the \"train2\" DataFrame\n",
    "drop_cols = [\n",
    "    \"predict_rows\",\n",
    "    'data_roc',\n",
    "    'period_of_day',\n",
    "    'data_roc_bin',\n",
    "]\n",
    "\n",
    "# Drop duplicates from the \"train2\" DataFrame\n",
    "train2.drop_duplicates(inplace=True)\n",
    "\n",
    "# Splitting the target variable and dropping columns from \"train2\" DataFrame\n",
    "y = train2[\"data_rate_t+1_trend\"]\n",
    "train2.drop(columns=[\"data_rate_t+1_trend\", \"NE ID\", \"endTime\"], inplace=True, axis=1)\n",
    "train2.drop(columns=drop_cols, inplace=True, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "61cf2390",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    482337\n",
       "0    464023\n",
       "Name: data_rate_t+1_trend, dtype: int64"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Count the occurrences of each unique value in the target variable \"y\"\n",
    "y.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "13966215",
   "metadata": {},
   "source": [
    "#### LightGBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "431582b6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] num_threads is set=8, n_jobs=1 will be ignored. Current value: num_threads=8\n",
      "1    1094\n",
      "0     838\n",
      "Name: data_rate_t+1_trend, dtype: int64\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>data_rate_t+1_trend</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>34572</th>\n",
       "      <td>B0017-25_24</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34574</th>\n",
       "      <td>B0017-25_25</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34575</th>\n",
       "      <td>B0017-25_26</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34578</th>\n",
       "      <td>B0017-25_27</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34725</th>\n",
       "      <td>B0017-32_1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                ID  data_rate_t+1_trend\n",
       "34572  B0017-25_24                    0\n",
       "34574  B0017-25_25                    0\n",
       "34575  B0017-25_26                    0\n",
       "34578  B0017-25_27                    1\n",
       "34725   B0017-32_1                    0"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Instantiate the LGBMClassifier model\n",
    "model = LGBMClassifier(random_state=seed,num_threads= 8,n_jobs=1)\n",
    "\n",
    "# Fit the model on the training data\n",
    "model.fit(train2, y)\n",
    "\n",
    "# Predict probabilities of class 1 for the test data using the trained model\n",
    "test_pred_probs = model.predict_proba(test[train2.columns])[:, 1]  # Probabilities of class 1\n",
    "\n",
    "# Adjust the threshold for classification based on probability values\n",
    "test_predictions = (test_pred_probs >= 0.49).astype(int)  # Adjust threshold as needed\n",
    "\n",
    "# Create a DataFrame to store predictions\n",
    "predictions_df = pd.DataFrame({'ID': test[\"NE ID\"], 'data_rate_t+1_trend': test_predictions})\n",
    "\n",
    "# Save the predictions to a CSV file\n",
    "predictions_df.to_csv(\"lgbm.csv\", index=False, header=True, sep=\",\", encoding='utf-8', index_label=False)\n",
    "\n",
    "# Display the value counts of predicted classes\n",
    "print(predictions_df[\"data_rate_t+1_trend\"].value_counts())\n",
    "\n",
    "# Display the first few rows of the predictions DataFrame\n",
    "predictions_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "e1a2062a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Get the feature importances from the trained LightGBM model\n",
    "feature_importances = model.feature_importances_\n",
    "\n",
    "# Get the names of the features from the X_train columns\n",
    "feature_names = train2.columns\n",
    "\n",
    "# Create a DataFrame to store feature names and their corresponding importances\n",
    "feature_importance_df = pd.DataFrame({'Feature': feature_names, 'Importance': feature_importances})\n",
    "\n",
    "# Sort the DataFrame based on feature importances in ascending order\n",
    "feature_importance_df = feature_importance_df.sort_values(by='Importance', ascending=True)\n",
    "\n",
    "# Plot the feature importances using matplotlib\n",
    "plt.figure(figsize=(10, 10))\n",
    "plt.barh(feature_importance_df['Feature'], feature_importance_df['Importance'])\n",
    "plt.xlabel('Feature Importance')\n",
    "plt.ylabel('Feature')\n",
    "plt.title('Feature Importances')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ec09702d",
   "metadata": {},
   "source": [
    "#### CatBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a129e1ca",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate the CatBoostClassifier model\n",
    "model = CatBoostClassifier(random_seed=seed, verbose=False,thread_count= 8)\n",
    "\n",
    "# Fit the model on the training data\n",
    "model.fit(train2, y)\n",
    "\n",
    "# Predict probabilities of class 1 for the test data using the trained model\n",
    "test_pred_proba = model.predict_proba(test[train2.columns])[:, 1]\n",
    "\n",
    "# Adjust the threshold for classification based on probability values\n",
    "test_predictions = (test_pred_proba >= 0.482).astype(int)\n",
    "\n",
    "# Create a DataFrame to store predictions\n",
    "predictions_df = pd.DataFrame({'ID': test[\"NE ID\"], 'data_rate_t+1_trend': test_predictions})\n",
    "\n",
    "# Save the predictions to a CSV file named \"catboost.csv\"\n",
    "predictions_df.to_csv(\"catboost.csv\", index=False)\n",
    "\n",
    "# Display the value counts of predicted classes\n",
    "print(predictions_df[\"data_rate_t+1_trend\"].value_counts())\n",
    "\n",
    "# Display the first few rows of the predictions DataFrame\n",
    "predictions_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d2accb17",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the feature importances from the trained LightGBM model\n",
    "feature_importances = model.feature_importances_\n",
    "\n",
    "# Get the names of the features from the X_train columns\n",
    "feature_names = train2.columns\n",
    "\n",
    "# Create a DataFrame to store feature names and their corresponding importances\n",
    "feature_importance_df = pd.DataFrame({'Feature': feature_names, 'Importance': feature_importances})\n",
    "\n",
    "# Sort the DataFrame based on feature importances in ascending order\n",
    "feature_importance_df = feature_importance_df.sort_values(by='Importance', ascending=True)\n",
    "\n",
    "# Plot the feature importances using matplotlib\n",
    "plt.figure(figsize=(10, 10))\n",
    "plt.barh(feature_importance_df['Feature'], feature_importance_df['Importance'])\n",
    "plt.xlabel('Feature Importance')\n",
    "plt.ylabel('Feature')\n",
    "plt.title('Feature Importances')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14359b98",
   "metadata": {},
   "source": [
    "#### Ensembling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e2339be6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read the model prediction CSV files\n",
    "catboost = pd.read_csv(\"catboost.csv\")\n",
    "lgbm = pd.read_csv(\"lgbm.csv\")\n",
    "\n",
    "# Create an empty DataFrame for the ensemble predictions\n",
    "ensemble = pd.DataFrame()\n",
    "\n",
    "# Copy the 'ID' column from the 'catboost' DataFrame to the 'ensemble' DataFrame\n",
    "ensemble[\"ID\"] = catboost[\"ID\"]\n",
    "\n",
    "# Concatenate the 'data_rate_t+1_trend' columns from both models into a single DataFrame\n",
    "predictions_data = pd.concat([catboost[\"data_rate_t+1_trend\"], lgbm[\"data_rate_t+1_trend\"]], axis=1)\n",
    "\n",
    "# Calculate the sum of predictions from both models for each row\n",
    "predictions_data[\"ensemble\"] = predictions_data.iloc[:, 0] + predictions_data.iloc[:, 1]\n",
    "\n",
    "# Update 'ensemble' column to 1 if the sum is greater than or equal to 1\n",
    "predictions_data.loc[predictions_data[\"ensemble\"] >= 1, \"ensemble\"] = 1\n",
    "\n",
    "# Assign the 'ensemble' column to the 'data_rate_t+1_trend' column in the 'ensemble' DataFrame\n",
    "ensemble[\"data_rate_t+1_trend\"] = predictions_data[\"ensemble\"]\n",
    "\n",
    "# Display the value counts of the 'data_rate_t+1_trend' column in the ensemble predictions\n",
    "print(ensemble[\"data_rate_t+1_trend\"].value_counts())\n",
    "\n",
    "# Save the ensemble predictions to a CSV file named 'ensemble.csv'\n",
    "ensemble.to_csv('ensemble.csv', index=False)\n",
    "\n",
    "# Display the first few rows of the ensemble predictions DataFrame\n",
    "ensemble.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "64ad5880",
   "metadata": {},
   "source": [
    "# END!!"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}