HAR_analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "Md8TSPPyJLy8"
   },
   "outputs": [],
   "source": [
    "# Necessary Imports\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "id": "ROGtrFPHJLy_"
   },
   "outputs": [],
   "source": [
    "# ML and Deep Learning Imports\n",
    "from sklearn.preprocessing import StandardScaler, MinMaxScaler\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score\n",
    "\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.losses import MeanSquaredError\n",
    "from tensorflow.keras.metrics import Accuracy\n",
    "\n",
    "from scipy.ndimage import gaussian_filter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "hdNVBvQDJdrg",
    "outputId": "33ff1ed5-69af-46a5-ed53-de68e28c8c58"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mounted at /content/gdrive\n"
     ]
    }
   ],
   "source": [
    "# Mounting my GoogleDrive, so that I can access the local data folders through the Colab environment\n",
    "from google.colab import drive\n",
    "drive.mount('/content/gdrive')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "d1ROyO50JLy3"
   },
   "source": [
    "------------------\n",
    "# **Human Acitivity Recognition Using Labeled Smartphone Data**\n",
    "\n",
    "### ***Project by Jonny Hofmeister***\n",
    "\n",
    "***Capstone Project for Flatiron School Data Science Bootcamp***\n",
    "\n",
    "This data was sourced from the UCI Machine Learning Repository - [\n",
    "Smartphone-Based Recognition of Human Activities and Postural Transitions Data Set](https://archive.ics.uci.edu/ml/datasets/Smartphone-Based+Recognition+of+Human+Activities+and+Postural+Transitions)\n",
    "\n",
    "The data was collected by Davide Anguita, Alessandro Ghio, Luca Oneto, Xavier Parra, and Jorge L. Reyes-Ortiz at the University of Genova. Their paper [A Public Domain Dataset for Human Activity Recognition Using Smartphones](https://www.esann.org/sites/default/files/proceedings/legacy/es2013-84.pdf) describes in-depth their process in collecting, processing, and labeling this data.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "euxLo0Y1JLy_"
   },
   "source": [
    "\n",
    "\n",
    "------------\n",
    "## **Business Understanding**\n",
    "\n",
    "The goal of this classification model is to use smartphone-based sensor data/information to determine what kind of activity the user is performing. There are 6 activities labeled and recorded in this data, three are static postures (Sitting, Standing, and Laying Down) and three dynamic activities (Walking, Walking Upstairs, and Walking Downstairs). The researchers have also labeled and included the transitions between these activities, but intial modeling attempts in this project will seek to classify just the initial 6 activities. \n",
    "\n",
    "Understanding the activities of persons can have large benefits towards understanding their health and daily patterns. Recognizing and recording these patterns can help determine how activites affect health, sleep, etc. For example, tracking time sitting down can help researches correlate time spent sitting to negative health effects and them begin to inform patients on how to change their activity patterns to improve health. \n",
    "\n",
    "Smartphones are something most people have on them most of the day, and therefore are a consistent way to record activity information. The downside, is that the location of the phone has a large influence on what the sensor data looks like for each activity. Having the phone in a back pocket vs in-hand while walking 'looks' different to the sensors and gyroscopes and thus limits our ability to properly classify the activity. In the long run, using data from consistently located sensors like smart-watches and fit-bands may provide more consistent results of a deployed model, but this readily-availble data from smartphones is a great place to begin research in creating models that can successfully classify activities. These researches recognized this location problem, and addressed it by consistently mounting the smartphone on the waist of all participants during the data collection process.\n",
    "\n",
    "The researchers state \"These mass-marketed devices provide a flexible, affordable and self-contained solution to automatically and unobtrusively monitor Activities of Daily Living\" as a benefit of using smartphones to collect this data."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kkXNcdl4JLzA"
   },
   "source": [
    "\n",
    "<img src='images/AdobeStock_367863903.jpeg' align=center width=700>\n",
    "(Image from Adobe Stock)\n",
    "\n",
    "--------------\n",
    "## **Data Understanding**\n",
    "\n",
    "The first step in understanding the data is to look at how it was collected. Conveniently, the researchers have provided [this video]('https://www.youtube.com/watch?v=XOEN9W05_4A') which shows an example of one of their collection trials."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mxIhJoc4JLzA"
   },
   "source": [
    "### **Understanding the Features**\n",
    "\n",
    "This dataset has been taken from the embedded accelerometer and gyroscopes in Samsung Galaxy II smartphones. The tri-axial (x, y, and z) signals were sampled a constant rate of 50Hz (50 samples per second). The sensor signals were then processed through noise filters and then sampled in 2.56 second sliding windows with 50% overlap - this corresponds to 128 readings per window.\n",
    "\n",
    "Then from each sample window, variables from the time and frequency domain were calculated to obtain a vector of 561 features. It is this 561 feature vector for each window that we use as our data samples to train a model.\n",
    "\n",
    "Lets explore exactly what these are. The time and frequency domain signals used to estimate the features are:\n",
    "\n",
    "    - tBodyAcc-XYZ\n",
    "    - tGravityAcc-XYZ\n",
    "    - tBodyAccJerk-XYZ\n",
    "    - tBodyGyro-XYZ\n",
    "    - tBodyGyroJerk-XYZ\n",
    "    - tBodyAccMag\n",
    "    - tGravityAccMag\n",
    "    - tBodyAccJerkMag\n",
    "    - tBodyGyroMag\n",
    "    - tBodyGyroJerkMag\n",
    "    - fBodyAcc-XYZ\n",
    "    - fBodyAccJerk-XYZ\n",
    "    - vfBodyGyro-XYZ\n",
    "    - fBodyAccMag\n",
    "    - fBodyAccJerkMag\n",
    "    - fBodyGyroMag\n",
    "    - fBodyGyroJerkMag\n",
    "    \n",
    "Where 'XYZ' denotes the tri-axial signals.\n",
    "\n",
    "Then to create the 561 feature vector, for each of the above signals, this set of variables was calcualted:\n",
    "\n",
    "    - mean(): Mean value\n",
    "    - std(): Standard deviation\n",
    "    - mad(): Median absolute deviation \n",
    "    - max(): Largest value in array\n",
    "    - min(): Smallest value in array\n",
    "    - sma(): Signal magnitude area\n",
    "    - energy(): Energy measure. Sum of the squares divided by the number of values. \n",
    "    - iqr(): Interquartile range \n",
    "    - entropy(): Signal entropy\n",
    "    - arCoeff(): Autorregresion coefficients with Burg order equal to 4\n",
    "    - correlation(): correlation coefficient between two signals\n",
    "    - maxInds(): index of the frequency component with largest magnitude\n",
    "    - meanFreq(): Weighted average of the frequency components to obtain a mean frequency\n",
    "    - skewness(): skewness of the frequency domain signal \n",
    "    - kurtosis(): kurtosis of the frequency domain signal \n",
    "    - bandsEnergy(): Energy of a frequency interval within the 64 bins of the FFT of each window.\n",
    "    - angle(): Angle between to vectors.\n",
    "    \n",
    "This is a lot of good data! As I mentioned in the above section, we don't want to manually perform feature selection from these 561 options and want to employ modeling techniques that do this for us. So, I use this entire dataset and all 561 features we have access to to train the models."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "6KMRw5IwJLzB"
   },
   "source": [
    "#### ***Test Subjects***\n",
    "\n",
    "The data was collected from 30 volunteers who all underwent the same process for data collection as outlined in the video. Becasue of the time-series nature of the data and inconsistent amounts of data for each test subject and activity, I plan on using the train/test split the researchers have used, in which they select 70% of volunteers as training data, and 30% for testing. This is slightly different than dividing the number of samples of data by the same ratios. I also will not apply any shuffling to the data before modeling - we want the data grouped by subject and then by activity because this is likely how real life (not just validation data) will be input into the model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bjvZmMf1JLzG"
   },
   "source": [
    "### **Data Exploration**\n",
    "\n",
    "#### ***Missing values***\n",
    "\n",
    "The first thing I did was search the training and testing sets for missing values. This was a very well constructed dataset and contained zero NaN or Null values and therefore we will not need to employ and kind of imputation step in the preprocessing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7lFfoa_JJLzH"
   },
   "source": [
    "#### ***Class Distribution***\n",
    "\n",
    "Next we explore the ditribution of the target classes in a bar plot shown below.\n",
    "\n",
    "<img align=\"center\" src=\"images/class_distribution.png\">\n",
    "\n",
    "While not perfectly balanced, the class distribution is fairly even. When considering the binary Dynamic vs Static groups, the imbalance is slightly greater. But given there is still a significant number of samples in the smaller class, I do not think it is necessary to weight or upsample one of the groups."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ZWPbuTJJJLzI"
   },
   "source": [
    "----------\n",
    "\n",
    "## **Data Preparation**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UGOF_5GXJLzI"
   },
   "source": [
    "Begin be extracting the feature names for the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "EswuCTBXJLzI"
   },
   "outputs": [],
   "source": [
    "# Open the features.txt file and read it into memory\n",
    "file = open('/content/gdrive/MyDrive/capstone/data/HAPT_data_set/features.txt', 'r')\n",
    "lines = file.readlines()\n",
    "\n",
    "# Take the readline strings and split it into just feature names\n",
    "columns = [line.replace('\\n', '').strip() for line in lines]\n",
    "file.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "1GJzcLelJLzI"
   },
   "source": [
    "Unfortunately, not all the feature names are unique, so we must first sort through and make adjustments to the dupplicates."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "_7JOKpafJLzJ",
    "outputId": "97d6d6d1-f59b-46ff-a77c-6bca2c47a1b6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "561"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# List to store feature names\n",
    "unique_cols = []\n",
    "\n",
    "# If a name is already in the unique cols list, we add it's index to the string so it becomes unique.\n",
    "for idx, val in enumerate(columns):\n",
    "    if val not in unique_cols:\n",
    "        unique_cols.append(val)\n",
    "    else:\n",
    "        unique_cols.append(val+'__'+str(idx))\n",
    "\n",
    "# Check the length\n",
    "len(unique_cols)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bS3G-aFPJLzL"
   },
   "source": [
    "Perfect, now we have all 561 feature names ready for importing the dataset. Below, I import the datafiles exactly as they are named and with the proper labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "1oeNzJsyJLzL"
   },
   "outputs": [],
   "source": [
    "# Importing the training set with the unique feature names and then the Activity labels for the training set.\n",
    "X_train = pd.read_csv('/content/gdrive/MyDrive/capstone/data/HAPT_data_set/Train/X_train.txt',\n",
    "                      delimiter=' ', engine='python', header=None, names=unique_cols)\n",
    "y_train = pd.read_csv('/content/gdrive/MyDrive/capstone/data/HAPT_data_set/Train/y_train.txt',\n",
    "                      header=None, names=['Activity_Label'])\n",
    "\n",
    "# Next, same thing for the Testing set.\n",
    "X_test = pd.read_csv('/content/gdrive/MyDrive/capstone/data/HAPT_data_set/Test/X_test.txt',\n",
    "                     delimiter=' ', engine='python', header=None, names=unique_cols)\n",
    "y_test = pd.read_csv('/content/gdrive/MyDrive/capstone/data/HAPT_data_set/Test/y_test.txt',\n",
    "                     header=None, names=['Activity_Label'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "IpBSWHoFJLzN"
   },
   "source": [
    "Before I relabel this dataset for the first and second stage classifiers, I am going to join the features df with the labels df. This is to create a simple and consistent single dataframe for the Training and Testing sets I can easily manupilate before inputting it into the first or second stage classifier model. \n",
    "\n",
    "Then, I add another label column for the binary classification group, either 'Dynamic' or 'Static'. The dynamic activities are Walking, Walking Upstairs, and Walking Downstairs, and are numbered activities 1, 2 and 3. The static activities are Sitting, Standing, and Laying, and are labeled 4, 5 and 6. Labels 7 though 12 are transition activities, i.e. Sitting to Standing. This model will not attempt to classify these transitions, so we remove row instances with activity labels greater than 6."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "30VtjGaIJLzN",
    "outputId": "a1b7c3be-3aea-40fe-e33a-b6b23c338bc1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set instances: 7415\n",
      "Testing set instances: 2996\n",
      "Train/Test Ratio: 71/28\n"
     ]
    }
   ],
   "source": [
    "# Join the labels and the features.\n",
    "train = y_train.join(X_train)\n",
    "test = y_test.join(X_test)\n",
    "\n",
    "# Remove transition activites\n",
    "train = train[train['Activity_Label'] < 7]\n",
    "test = test[test['Activity_Label'] < 7]\n",
    "\n",
    "# Add binary activity column for first stage classification\n",
    "train['binary_activity'] = np.where(train['Activity_Label'] < 4, 'D', 'S')\n",
    "test['binary_activity'] = np.where(test['Activity_Label'] < 4, 'D', 'S')\n",
    "\n",
    "# Check the lengths and that the train test split ratio was retained after removing transition activities.\n",
    "print(f'''Training set instances: {len(train)}\n",
    "Testing set instances: {len(test)}\n",
    "Train/Test Ratio: {int(len(train)/(len(train)+len(test))*100)}/{int(len(test)/(len(train)+len(test))*100)}''')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Hgi9RVqtJLzN"
   },
   "source": [
    "Awesome, now we have a Dataframe set up for the training and testing sets that can be easily subset for input into either of the three models we need to train (the first stage binary, and two second stage 3-class models). \n",
    "\n",
    "The training and testing split is just about 70/30. This corresponds to the split performed by the researchers. The data was split by subject, not random rows. We dont have exactly 70/30 because of slight variances in the subject's trials and the time spent in transition activities, which we removed. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jrDXTK5YJLzO"
   },
   "source": [
    "#### ***Scaling***\n",
    "\n",
    "I do not apply any scaling or normalization techniques in this model. As the raw data has been processed and extracted for statistical features already, all features are in the range from -1 to 1. Also, the goal of this model isnt to extract information comparing the effect sizes of the different features. We want the model to automatically perform feature selection so we dont have to. For these reasons, mainly that the data is mostly normal and scaled to -1 to 1 already, I do not apply any scaling during the preprocessing."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tdRzRcqtoW8t"
   },
   "source": [
    "#### ***Data Sharpening***\n",
    "\n",
    "The final step of data preparation is to apply sharpening techniques to the features. But, we only want to sharpen the data for the 2nd stage of the model. So, I leave our 'train' and 'test' dataframes un-sharpened and apply the filters as we subset and construct the specific training and testing sets to for input to each of the three models.\n",
    "\n",
    "----------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Zs_ndodvJLzO"
   },
   "source": [
    "## **Modeling**\n",
    "\n",
    "### **Process**\n",
    "\n",
    "For this proof of concept HAR analysis, I employ the \"Divide and Conquor\" method proposed in [this research paper](https://www.mdpi.com/1424-8220/18/4/1055).\n",
    "\n",
    "<img align=\"center\" src=\"images/2_stage_classification.png\" width=\"650\">\n",
    "\n",
    "*Credit for this image belongs to the authors of the Divide and Conquor Paper.*\n",
    "\n",
    "This method breaks modeling up into 2 stages. First, we perform a binary classification that groups the data into Dynamic and Static activities. Then second, we perform two 3-class classifications for the 3 activities belonging to the dynamic and static groups. Walking, walking upstairs, and walking downstairs are dynamic activities. Sitting, standing, and laying are static activities.\n",
    "\n",
    "### **First-Stage Binary Classifier**\n",
    "\n",
    "The researchers who first proposed this method found that a decision tree was able to classify Dynamic and Static acivities with a perfect accuracy. Because of the known success of this method and the simplicity of decision trees/ random forests, I choose to implement one for the binary stage of my model.\n",
    "\n",
    "While the researchers used a decision tree, I will implement a random forest because they can slightly reduce over-fitting and they do not take any more effort to employ than a decision tree. I will utilize the Sci-Kit Learn package to build this model.\n",
    "\n",
    "We mentioned before that with so many features in our data, a model that can automatically perform feature selection is ideal. This is a strength of decision trees and random forests because each branch 'decision' is based off of feature conditions that are most important and influential. This makes our model quickly adaptable to new datasets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "FBeKt8w2JLzO",
    "outputId": "615504f4-3d76-4454-edcc-94807107ed25"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(n_estimators=40, random_state=1234)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Begin my selecing the data and proper labels to train the first stage binary with.\n",
    "X_train_binary = train.drop(['Activity_Label', 'binary_activity'], axis=1)\n",
    "y_train_binary = train['binary_activity']\n",
    "\n",
    "X_test_binary = test.drop(['Activity_Label', 'binary_activity'], axis=1)\n",
    "y_test_binary = test['binary_activity']\n",
    "\n",
    "\n",
    "# Now build the Random Forest Classifier\n",
    "rf_clf = RandomForestClassifier(random_state=1234,\n",
    "                                n_estimators=40)\n",
    "\n",
    "# Fit the RF on the training set\n",
    "rf_clf.fit(X_train_binary, y_train_binary)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jGDu6Y6lJLzO"
   },
   "source": [
    "While training this model for the first time, the training and testing sets only had a couple incorrect predictions. I was able to balance this by adjusting only one hyperparameter - the number of estimators. A higher number of estimators (n=100) slightly overfit and would get a few of the test predictions wrong. A lower number of estimators (n=10) yeilded some incorrect guessed on the training set but was able to generalize the smaller testing set perfectly. After a few tries, I found that a middle ground of n_estimators=40 was able to balance generalization and overfitting and classify both training and testing perfectly."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-eD5eeqTJLzO"
   },
   "source": [
    "### **Second-Stage 3-Class Models**\n",
    "\n",
    "The two second stage models will both utilize neural networks and the Keras package. Both models will employ 1-dimensional convolutional methods/layers but will be structured slightly differently. I expereimented with slightly different constructions in these models, but they are both based off the second stage 1D CNN models proposed by [this HAR research paper]('https://www.mdpi.com/1424-8220/18/4/1055') I have previously mentioned.\n",
    "\n",
    "### **\"Dynamic\" Activity 1-D Convolutional Neural Network**\n",
    "\n",
    "The first thing that needs to be done is to select data samples only for dynamic activities. Walking, Walking Upstairs, and Walking Downstairs are labels 1, 2 and 3.  Next we split the training and testing sets into into the features (X) and labels (y). Since this is a 3 class model and not binary, we must One Hot Encode the target labels. We will do this using Keras to_categorical utility, which requires our labels begin at 0. To quickly fix this, we replace label class '3' with '0' and leave 1 and 2 unaltered.\n",
    "\n",
    "Then we put this formatted data back into dataframes before our scaling step. For clarity, Walking Downstairs label is 0, Walking is 1, and Walking Upstairs is 2."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "aOUg-BQslMkA",
    "outputId": "db88a699-098e-4e29-c87e-80f4c453ba2e"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/pandas/core/series.py:4582: 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",
      "  method=method,\n"
     ]
    }
   ],
   "source": [
    "# Selecting just the dynamic activity data from training and testing sets\n",
    "dynamic_train = train[train['Activity_Label'] <= 3]\n",
    "dynamic_test = test[test['Activity_Label'] <= 3]\n",
    "\n",
    "# Splitting into data and labels\n",
    "# Training\n",
    "X_train_dynamic = dynamic_train.drop(['Activity_Label', 'binary_activity'], axis=1)\n",
    "y_train_dynamic = dynamic_train['Activity_Label']\n",
    "# Testing\n",
    "X_test_dynamic = dynamic_test.drop(['Activity_Label', 'binary_activity'], axis=1)\n",
    "y_test_dynamic = dynamic_test['Activity_Label']\n",
    "\n",
    "# The Keras util to_categorical requires the integers for labels begin at 0, the labels in this set are currently 1,2,3.\n",
    "y_train_dynamic.replace(3, 0, inplace=True)\n",
    "y_test_dynamic.replace(3, 0, inplace=True)\n",
    "\n",
    "# Convert targets to OHE categorical\n",
    "y_train_dynamic = to_categorical(y_train_dynamic)\n",
    "y_test_dynamic = to_categorical(y_test_dynamic)\n",
    "# And back to a pandas DataFrame\n",
    "y_train_dynamic = pd.DataFrame(y_train_dynamic)\n",
    "y_test_dynamic = pd.DataFrame(y_test_dynamic)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "isvkPKR5fq3X"
   },
   "source": [
    "#### ***Sharpening***\n",
    "\n",
    "Applying data sharpening before the second stage models was found to increase performance by the research paper. Their proposed method for sharpening is as follows:\n",
    "  - Denoise the data with a gaussian filter.\n",
    "  - Exctract details by subtracting the denoised result from the data.\n",
    "  - Amplifying these details to some degree using a chosen parameter,\n",
    "  - Then add these details back to the original data to obtain the sharpened result.\n",
    "\n",
    "As mentioned before, we apply sharpening now as we do not want to use sharpened data in our 1st stage decision tree. We also sharpen separately for Dynamic and Static models so that we can slightly adjust the denoising and amplification amounts to best fit the differences in dynamic and static activity data.\n",
    "\n",
    "In this case, the dynamic model applies a higher degree of detail amplification in order to produce higher contrast between activity signals than in the static model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "9NKipcs-cwW1"
   },
   "outputs": [],
   "source": [
    "# Data sharpening\n",
    "# Begin be denoising data with gaussian filter\n",
    "sigma = 7\n",
    "denoised_X_train_dynamic = gaussian_filter(X_train_dynamic, sigma=sigma)\n",
    "denoised_X_test_dynamic = gaussian_filter(X_test_dynamic, sigma=sigma)\n",
    "\n",
    "# The denoised result is subtracted from the test data to obtain sharped details\n",
    "detailed_X_train_dynamic = X_train_dynamic - denoised_X_train_dynamic\n",
    "detailed_X_test_dynamic = X_test_dynamic - denoised_X_test_dynamic\n",
    "\n",
    "# The sharpened details are then amplified to some degree using α parameter and added to the original test data to obtain sharpened test data\n",
    "alpha = 1.95\n",
    "sharpened_X_train_dynamic = X_train_dynamic + (alpha * detailed_X_train_dynamic)\n",
    "sharpened_X_test_dynamic = X_test_dynamic + (alpha * detailed_X_test_dynamic)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "2lC8nsmT9f9o",
    "outputId": "9363c043-3b5f-47a1-f998-e2365d61487f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3286, 561)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check the input shape\n",
    "sharpened_X_train_dynamic.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "q_drmdpdRJ3d"
   },
   "source": [
    "Great, the features have been sharpened and targets encoded. We have 3286 samples in the Dynamic training set and the input dimensions for the neural network will be (561, 1) as we have 561 features that are all a single value.\n",
    "\n",
    "##### ***CNN Construction***\n",
    "\n",
    "The Dynamic activity classification model is a 1-D convolutional neural network. We utilize a single 1D convolution layer followed by a Max pooling layer. The kernel size of the convolution is set to the stride size for the pooling layer. \"ReLu\" is the activation function we use for the convolutional layer. \n",
    "\n",
    "After the convolutional step, we flatten the nueral network, apply a dropout layer at a rate of 0.5 to help reduce overfitting, and finally a output layer with 'softmax' activation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "id": "pObpbMTjJLzP"
   },
   "outputs": [],
   "source": [
    "\n",
    "# Building the \"Dynamic\" CNN Classifier\n",
    "dynamic_cnn = Sequential([\n",
    "  # Begin with the 1D Convolution, window and stride of 3\n",
    "  layers.Conv1D(100, kernel_size=3, input_shape=(561, 1), activation='relu'),\n",
    "  # Pooling Layer\n",
    "  layers.MaxPooling1D(strides=3),\n",
    "  # Flattening layer\n",
    "  layers.Flatten(),\n",
    "  # Dense layer with 50% dropout\n",
    "  layers.Dropout(.5),\n",
    "  # Output layer, 3-class softmax\n",
    "  layers.Dense(3, activation='softmax')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "tz68X31yJLzP",
    "outputId": "4be93b4d-9c7d-42df-ed70-8838073d0f8f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv1d_5 (Conv1D)           (None, 559, 100)          400       \n",
      "                                                                 \n",
      " max_pooling1d_2 (MaxPooling  (None, 186, 100)         0         \n",
      " 1D)                                                             \n",
      "                                                                 \n",
      " flatten_3 (Flatten)         (None, 18600)             0         \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 18600)             0         \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 3)                 55803     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 56,203\n",
      "Trainable params: 56,203\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "dynamic_cnn.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tTnGTPA0nuVf"
   },
   "source": [
    "56 thousand trainable parameters is relatively low, so this model will train quickly. We select the Adam optimizer, MSE as our loss function, and score the model on accuracy.\n",
    "\n",
    "We use the standard batch size of 32. During intial exploration, this model learned, but slowly. So, we set the number of epochs very high, 500, to give the model time to improve. This is feasible because each epoch trains in only 4 seconds."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "vVvaCwyJs_vV"
   },
   "outputs": [],
   "source": [
    "dynamic_cnn.compile(optimizer='adam',\n",
    "                    loss=MeanSquaredError(),\n",
    "                    metrics=[Accuracy()])\n",
    "\n",
    "epochs = 500\n",
    "\n",
    "dynamic_cnn_results = dynamic_cnn.fit(\n",
    "    x = sharpened_X_train_dynamic, \n",
    "    y = y_train_dynamic,\n",
    "    validation_data=[sharpened_X_test_dynamic, y_test_dynamic],\n",
    "    epochs = epochs,\n",
    "    batch_size = 32\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ACtTdMjfsFTe"
   },
   "source": [
    "The model has been trained and results stored. So that we don't have to re-run the model to extract results, the below cell includes code to save and import the model. Note that this code is specific for my Google Colab environment."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "id": "K4wzjW9vsGBT"
   },
   "outputs": [],
   "source": [
    "# Save model\n",
    "# dynamic_cnn.save('/content/gdrive/MyDrive/capstone/models/dynamic_cnn')\n",
    "\n",
    "# Load model\n",
    "# dynamic_cnn = keras.models.load_model('/content/gdrive/MyDrive/capstone/models/dynamic_cnn')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-wGSDIgDJLzP"
   },
   "source": [
    "### **\"Static\" Activity 1-D Convolutional Neural Network**\n",
    "\n",
    "Begin by selecting the static activity samples from the training and testing sets. Same exact process as above with the dynamic model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "1n--zq1DJLzP",
    "outputId": "430fae1c-d263-4960-da40-666a5dc1e105"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/pandas/core/series.py:4582: 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",
      "  method=method,\n"
     ]
    }
   ],
   "source": [
    "# Selecting just the static activity data from training and testing sets\n",
    "static_train = train[train['Activity_Label'] > 3]\n",
    "static_test = test[test['Activity_Label'] > 3]\n",
    "\n",
    "# Splitting into data and labels\n",
    "# Training\n",
    "X_train_static = static_train.drop(['Activity_Label', 'binary_activity'], axis=1)\n",
    "y_train_static = static_train['Activity_Label']\n",
    "# Testing\n",
    "X_test_static = static_test.drop(['Activity_Label', 'binary_activity'], axis=1)\n",
    "y_test_static = static_test['Activity_Label']\n",
    "\n",
    "# The Keras util to_categorical requires the integers for labels begin at 0, the labels in this set are currently 1,2,3.\n",
    "y_train_static.replace([4, 5, 6], [0, 1, 2], inplace=True)\n",
    "y_test_static.replace([4, 5, 6], [0, 1, 2], inplace=True)\n",
    "\n",
    "# Convert targets to OHE categorical\n",
    "y_train_static = to_categorical(y_train_static)\n",
    "y_test_static = to_categorical(y_test_static)\n",
    "# And back to a pandas DataFrame\n",
    "y_train_static = pd.DataFrame(y_train_static)\n",
    "y_test_static = pd.DataFrame(y_test_static)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "whfiQ374Ywdf"
   },
   "source": [
    "#### ***Sharpening***\n",
    "\n",
    "Next, the data is sharpened with the same process as above. The only differences are the sigma and alpha parameters used in the gaussian filter and amplification. The alpha value in this case does not amplify the details nearly as much as for the previous model. These parameters are suggested by the researchers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "4QW6jcsVxpb0"
   },
   "outputs": [],
   "source": [
    "# Data sharpening\n",
    "# Begin be denoising data with gaussian filter\n",
    "sigma = 8\n",
    "denoised_X_train_static = gaussian_filter(X_train_static, sigma=sigma)\n",
    "denoised_X_test_static = gaussian_filter(X_test_static, sigma=sigma)\n",
    "\n",
    "# The denoised result is subtracted from the test data to obtain sharped details\n",
    "detailed_X_train_static = X_train_static - denoised_X_train_static\n",
    "detailed_X_test_static = X_test_static - denoised_X_test_static\n",
    "\n",
    "# The sharpened details are then amplified to some degree using α parameter and added to the original test data to obtain sharpened test data\n",
    "alpha = 0.07\n",
    "sharpened_X_train_static = X_train_static + (alpha * detailed_X_train_static)\n",
    "sharpened_X_test_static = X_test_static + (alpha * detailed_X_test_static)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "Vdcns7G2xpS4",
    "outputId": "1b4e45b4-4b10-477a-8b92-4c3210b2a416"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4129, 561)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check the input shape\n",
    "sharpened_X_train_static.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "q6grORH5ZjPd"
   },
   "source": [
    "The static model contains more training samples than the dynamic set, almost a thousand more. This will benefit performance and slightly reduce training time. We still have the same (561, 1) features input dimension for the neural network.\n",
    "\n",
    "***Static Model Construction*** \n",
    "\n",
    "The 1D CNN for the static activity classification is slightly different. Rather than utilizing a Max Pooling layer, we employ 3 consecutive 1D convolution layers with an increasing number of filters. The window size and activation function used for the Conv layers is still 3 and 'ReLu' respectively. The second half of this model matches the dynamic classifier with a flattening layer, a dense/dropout layer set to 0.5 dropout, and a 3 class 'softmax' output layer. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "id": "T10PXDcxyRTR"
   },
   "outputs": [],
   "source": [
    "# Building the \"Dynamic\" CNN Classifier\n",
    "static_cnn = Sequential([\n",
    "  # Static model contains 3 progressive 1D convolution layers\n",
    "  layers.Conv1D(30, kernel_size=3, input_shape=(561, 1), activation='relu'),\n",
    "  layers.Conv1D(50, kernel_size=3, activation='relu'),\n",
    "  layers.Conv1D(100, kernel_size=3, activation='relu'),\n",
    "  # Flatten layer\n",
    "  layers.Flatten(),\n",
    "  # Dense layer with 50% dropout\n",
    "  layers.Dropout(.5),\n",
    "  # Output layer, 3-class softmax\n",
    "  layers.Dense(3, activation='softmax')\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xVOUGr9PyRE1",
    "outputId": "d7053e5c-0b01-4cce-c9da-11c9f6463cb7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv1d_1 (Conv1D)           (None, 559, 30)           120       \n",
      "                                                                 \n",
      " conv1d_2 (Conv1D)           (None, 557, 50)           4550      \n",
      "                                                                 \n",
      " conv1d_3 (Conv1D)           (None, 555, 100)          15100     \n",
      "                                                                 \n",
      " flatten_1 (Flatten)         (None, 55500)             0         \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 55500)             0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 3)                 166503    \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 186,273\n",
      "Trainable params: 186,273\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "static_cnn.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "UkbD-COTxpJa"
   },
   "outputs": [],
   "source": [
    "static_cnn.compile(optimizer='adam',\n",
    "                    loss=MeanSquaredError(),\n",
    "                    metrics=[Accuracy()])\n",
    "\n",
    "epochs = 444\n",
    "\n",
    "static_cnn_results = static_cnn.fit(\n",
    "    x = sharpened_X_train_static, \n",
    "    y = y_train_static,\n",
    "    validation_data=[sharpened_X_test_static, y_test_static],\n",
    "    epochs = epochs,\n",
    "    batch_size = 32\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "_8mZGncJtRNP"
   },
   "outputs": [],
   "source": [
    "# Save model\n",
    "# static_cnn.save('/content/gdrive/MyDrive/capstone/models/static_cnn')\n",
    "\n",
    "# Load model\n",
    "# static_cnn = keras.models.load_model('/content/gdrive/MyDrive/capstone/models/static_cnn')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uB2lsaYSJLzP"
   },
   "source": [
    "## **Evaluation**\n",
    "------"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vCC84vp6JLzP"
   },
   "source": [
    "### **First Stage Binary Prediction Evaluation**\n",
    "\n",
    "To evaluate the first stage model we need to check performance on the training and testing sets. For the training set, I generate predictions and compute and accuracy score based on the true labels."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "na-jDSMJJLzP",
    "outputId": "d45bee2e-a32f-444f-b188-c9b95112e33f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1st Stage Training Accuracy: 1.0\n"
     ]
    }
   ],
   "source": [
    "# Checking training set accuracy\n",
    "train_preds_binary = rf_clf.predict(X_train_binary)\n",
    "print(f'1st Stage Training Accuracy: {accuracy_score(y_train_binary, train_preds_binary)}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xTI3tjWYJLzQ"
   },
   "source": [
    "Perfect score on the training set! Next let's see how well the model generalized to the testing data. Below I plot a confusion matrix to show a better visual of the results than just a score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 444
    },
    "id": "3_qTrGSwJLzQ",
    "outputId": "58c97e20-49d0-4f5a-d38b-dfe550b4eb87"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 648x504 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Generate predictions from the test set\n",
    "binary_preds = rf_clf.predict(X_test_binary)\n",
    "\n",
    "# Get the confusion matrix\n",
    "cf_matrix = confusion_matrix(y_test_binary, binary_preds)\n",
    "\n",
    "# Make it pretty with seaborn\n",
    "fig, ax = plt.subplots(figsize=(9, 7))\n",
    "\n",
    "sns.heatmap(cf_matrix, annot=True, fmt=\"d\", ax=ax, cmap='flare_r')\n",
    "ax.set_title('First Stage Confusion Matrix');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "k1yIjmA-JLzQ"
   },
   "source": [
    "Awesome, we perfectly classified the testing set too! Now that we have successfully grouped the activities into Dynamic and Static groups, we have eliminated a lot of error and uncertainty for the second stage models. Time to see how those performed."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ry2x2GbmJLzQ"
   },
   "source": [
    "### **Second Stage 1D CNN \"DYNAMIC Activity\" Model Evaluation**\n",
    "\n",
    "To evaluate the training and performace of our CNNs, let's plot the evolution of the model accuracy and loss as we progressed through training epochs. The loss function we employed was mean squared error, and accuracy is simply the success rate for all predictions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 390
    },
    "id": "e8VH4V6AxoEz",
    "outputId": "5b550abc-ba06-428a-962e-9495924eb03d"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dy_acc = dynamic_cnn_results.history['accuracy']\n",
    "dy_val_acc = dynamic_cnn_results.history['val_accuracy']\n",
    "\n",
    "dy_loss = dynamic_cnn_results.history['loss']\n",
    "dy_val_loss = dynamic_cnn_results.history['val_loss']\n",
    "\n",
    "epochs_range = range(epochs)\n",
    "\n",
    "plt.figure(figsize=(16, 6))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(epochs_range, dy_acc, label='Training Accuracy')\n",
    "plt.plot(epochs_range, dy_val_acc, label='Validation Accuracy')\n",
    "plt.legend(loc='lower right')\n",
    "plt.title('Training and Validation Accuracy')\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(epochs_range, dy_loss, label='Training Loss')\n",
    "plt.plot(epochs_range, dy_val_loss, label='Validation Loss')\n",
    "plt.legend(loc='upper right')\n",
    "plt.title('Training and Validation Loss');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GpJSYPFUAUOi"
   },
   "source": [
    "The dynamic activity classification scored around 80% accuracy for the training set and about 65% accuracy for the testing set. This isn't bad, and it appears from the accuracy score that the model had not quite plateaued and could possibly learn even more from the data. Another thing to notice is that the learning patterns for training and testing matched well. For example, around the 210th epoch, the accuracy scores in both sets started to trend upwards more quickly. This implies that patterns found in the training set are generalizing to the testing data very well - this is awesome."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2cFHf0SJJLzQ"
   },
   "source": [
    "### **Second Stage 1D CNN \"STATIC Activity\" Model Evaluation**\n",
    "\n",
    "Next, let's perform the same evaluation for the Static activity model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 390
    },
    "id": "eZpyRyiqLzz4",
    "outputId": "0df34779-ab71-42ea-e1ed-57c097cb2ee7"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x432 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "st_acc = static_cnn_results.history['accuracy']\n",
    "st_val_acc = static_cnn_results.history['val_accuracy']\n",
    "\n",
    "st_loss = static_cnn_results.history['loss']\n",
    "st_val_loss = static_cnn_results.history['val_loss']\n",
    "\n",
    "epochs_range = range(epochs)\n",
    "\n",
    "plt.figure(figsize=(16, 6))\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.plot(epochs_range, st_acc, label='Training Accuracy')\n",
    "plt.plot(epochs_range, st_val_acc, label='Validation Accuracy')\n",
    "plt.legend(loc='lower right')\n",
    "plt.title('Training and Validation Accuracy')\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.plot(epochs_range, st_loss, label='Training Loss')\n",
    "plt.plot(epochs_range, st_val_loss, label='Validation Loss')\n",
    "plt.legend(loc='upper right')\n",
    "plt.title('Training and Validation Loss');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "N_bfFNaABNqy"
   },
   "source": [
    "The Static activity model did excellent, training and testing scores were both above 95% in the later epochs. Loss scores for testing were slightly higher than training as expected, but both still very low. \n",
    "\n",
    "As in the dynamic model, we see a correlation in learning (accuracy score increase) in the training and testing sets. This implies the patterns learned in training are also found in the testing set and that the model should be fairly generalizable and robust. Overall, I am very happy with the result in this model and it is certainly good enough to move forward with."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "piILTDh2rOtb"
   },
   "source": [
    "### **Overall Model Evaluation**\n",
    "\n",
    "To visualize the results of the end-to-end model, we will return the predictions to their original labels, and concatenate the dynamic and static sets. Fortunately this is easy to do because the first stage model had 100% accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "SW7lvowRXu5M",
    "outputId": "529ed131-c655-4916-f79c-fcc6212de6ba"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Start-to-finish Confusion Matrix - Testing set\n",
    "\n",
    "# Make a dictionary for the integer labels and activity classes\n",
    "# Note I retain Walking upstairs as 0 (instead of 3), but change the static labels back to 4,5,6.\n",
    "activity_dict = {0: 'WU', 1:'W', 2:'WD', 4:'SIT', 5:'STAND', 6:'LAY'}\n",
    "\n",
    "# Test predictions from Dynamic model\n",
    "dynamic_preds = dynamic_cnn.predict(sharpened_X_test_dynamic)\n",
    "dynamic_preds = dynamic_preds.argmax(axis=-1)\n",
    "\n",
    "static_preds = static_cnn.predict(sharpened_X_test_static)\n",
    "static_preds = pd.DataFrame(static_preds.argmax(axis=-1)) # turn into DF so we can use df.replace() method\n",
    "# Convert static numeric labels back to 4, 5, 6 so we can tell them apart from dynamic\n",
    "static_preds.replace([0,1,2], [4,5,6], inplace=True)\n",
    "static_preds = np.transpose(np.asarray(static_preds))[0] # turn back into an array\n",
    "\n",
    "# Preds into single array\n",
    "preds = np.append(dynamic_preds, static_preds)\n",
    "\n",
    "# Now gather the true labels\n",
    "true_dy = np.asarray(y_test_dynamic).argmax(axis=-1)\n",
    "true_st = pd.DataFrame(np.asarray(y_test_static).argmax(axis=-1))\n",
    "true_st.replace([0,1,2], [4,5,6], inplace=True)\n",
    "# True into single array\n",
    "true = np.append(true_dy, true_st)\n",
    "\n",
    "# Check the lengths match\n",
    "len(preds) == len(true)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 729
    },
    "id": "oSo0HJ43wZDe",
    "outputId": "5676aef7-10e0-4b0c-fac9-2c9538296cd2"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Now plot the complete confusion matrix\n",
    "cf_matrix = confusion_matrix(true, preds)\n",
    "\n",
    "# Define the proper labels - I determined the order by manually checking counts\n",
    "labels = ['WALK UP', 'WALK', 'WALK DOWN', 'SIT', 'STAND', 'LAY']\n",
    "\n",
    "# Make it pretty with seaborn\n",
    "fig, ax = plt.subplots(figsize=(10, 10))\n",
    "\n",
    "sns.heatmap(cf_matrix, annot=True, \n",
    "            xticklabels=labels, yticklabels=labels,\n",
    "            fmt=\"d\", ax=ax, cmap=sns.cubehelix_palette(start=.5, rot=-.75, as_cmap=True),\n",
    "            linewidths=0.5, linecolor='w', cbar=False)\n",
    "ax.set_title('End-to-End HAR Confusion Matrix', fontdict={'fontsize': 20, 'weight': 'bold'})\n",
    "ax.set_xlabel('Predicted', fontdict={'fontsize': 15, 'weight': 'heavy'})\n",
    "ax.set_ylabel('True', fontdict={'fontsize': 15, 'weight': 'bold'})\n",
    "plt.tight_layout();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "40K9N_T-87_j"
   },
   "source": [
    "These are the final results of the end to end 2-stage HAR prediction model. One can finally see the strength of the 1st stage binary classifier here by examining the lower left and upper right quadrants of the confusion matrix - they are all zeros. This makes sure we never mis-classify a dynamic activity as static or vice versa. This step will also significantly increase the overall accuracy of the model.\n",
    "\n",
    "The most mis-classified activity was sitting. This model predicting standing 42 times when the true label was sitting. The second most mixed-up activity was predicting walking when the true label was walking down. 25 and 42 mis classifications are very insignificant given thousands of samples. Also, it is important to note it might not be as bad to mix up two dynamic or two static activites as it might be to mix up a dynamic activity for a static one. This is another strength of a strong first stage model.\n",
    "\n",
    "Lets's figure out what the actual accuracy score for the model is below."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "MZdbbbAFzhBO",
    "outputId": "2d4e8d31-542e-420d-dfaf-a9333ea6b6a2"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Overall Model Accuracy Score: 0.97\n"
     ]
    }
   ],
   "source": [
    "# Compute overall accuracy score\n",
    "print(f'Overall Model Accuracy Score: {accuracy_score(true, preds):.2f}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HYZtmcyK-NBY"
   },
   "source": [
    "An overall accuracy of 97% is awesome for a proof of concept analysis. I think the strength of this model comes from a first stage model that can precisely split the data by dynamic and static activities. This eliminates a huge number of misclassification possiblities and allows for more precise/specific models in the second stage. \n",
    "\n",
    "#### Sharpening Success\n",
    "\n",
    "I must also directly note the value of adding the data sharpening techniques proposed by the research paper. I attempted modeling without sharpening, and was unable to acheive accuracy scores above 45% in the dynamic 2nd stage model. The sharpening step is essential to the second stage models and also allows us to tweak the models more precisely."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "S9wi5DP9JLzQ"
   },
   "source": [
    "## **Next Steps**\n",
    "\n",
    "### **Improvements**\n",
    "\n",
    "The Binary and Static models acheived perfect and near perfect accuracy scores. But the Dynamic model was slightly lacking compared to its siblings. I think moving forward, one of the first things to do is tweak the dynamic model until it performs as well as its second stage static counterpart.\n",
    "\n",
    "Improving the dynamic model could be done by changing the structure and layers of the 1-D CNN, or by adjusting the parameters involved in the data sharpening step. The research paper outlines many possibilites of combinations of successful sharpening parameters, and trying these would be my first step before adjusting the CNN structure. \n",
    "\n",
    "We also had almost 1000 more training samples in the static model than in the dyanmic model. This also has an effect on how much the model can learn. So improving the dynamic model scores could also come from simply aquiring more data.\n",
    "\n",
    "### **Streamlining**\n",
    "\n",
    "Since this model so far is a proof of concept for a 2 stage model, it is broken up into 3 different models that must be run separately. Another next step now that we have created good enough models is to pipe them all together end-to-end. While improving and exploring models might be best done on each model separately as I have done here, testing this model on new data or deploying it would be much easier if it were all piped into a single function/model.\n",
    "\n",
    "### **Deployment Details**\n",
    "\n",
    "Also within the context of deployment, we next need to consider how data will be input and output and what information we want to extract from the model.\n",
    "\n",
    "Here, single row instances are classified into activity. In practice, we are more likely to input many consecutive samples/rows in time into the model. This would then output a prediction for activity for each window in time. We would then have to decide if we wanted the model to return the most likely or frequent activity for all the data input, or a breakdown of the different activities used. One example for a minute of sensor data could be to classify all rows, choose the most likely activity for each second, and then return information about how much time the user spent doing walking or doing a static activity. This is the kind of information that then becomes actionable to the user and tells them about their activity patterns through the day and how it relates to other health information like heart rate or blood pressure.\n",
    "\n",
    "<img src='images/AdobeStock_122055934.jpeg' align=center width=600>\n",
    "Image from Adobe Stock."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ag3ZhPaSlwH-"
   },
   "source": [
    "### **Resources / References**\n",
    "Data:\n",
    "  - [UCI](https://archive.ics.uci.edu/ml/datasets/Smartphone-Based+Recognition+of+Human+Activities+and+Postural+Transitions)\n",
    "  - [Data Aquisition Paper](https://www.esann.org/sites/default/files/proceedings/legacy/es2013-84.pdf)\n",
    "\n",
    "Tools / Packages:\n",
    "  - [Tensorflow](https://www.tensorflow.org/api_docs/python/tf)\n",
    "  - [Sci-Kit Learn](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html?highlight=random%20forest#sklearn.ensemble.RandomForestClassifier)\n",
    "\n",
    "References / Sources:\n",
    "  - [Divide and Conquor HAR Research paper](https://www.mdpi.com/1424-8220/18/4/1055)\n",
    "  - [Machine Learning Mastery - Summary of the different HAR research methods](https://machinelearningmastery.com/deep-learning-models-for-human-activity-recognition/)\n",
    "  - [ML Mastery - 1D CNN for HAR](https://machinelearningmastery.com/cnn-models-for-human-activity-recognition-time-series-classification/)\n",
    "\n",
    "\n",
    "\n",
    "  -----\n",
    "\n",
    "\n",
    "\n",
    "**Analysis by Jonny Hofmeister. Thanks for reading!**\n",
    "\n",
    "For questions, please contact me at jonny.hofmeister@gmail.com"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "HAR_analysis.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "learn-env",
   "language": "python",
   "name": "learn-env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}