HW9_Regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Линейная регрессия\n",
    "__Суммарное количество баллов: 10__\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import make_blobs, make_moons\n",
    "from sklearn.model_selection import train_test_split\n",
    "import numpy as np\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib\n",
    "import copy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_data(path=\"./hw9_data/boston.csv\"):\n",
    "    dataframe = np.genfromtxt(path, delimiter=\",\", skip_header=15)\n",
    "    X = dataframe[:, :-1]\n",
    "    y = dataframe[:, -1]\n",
    "    return X, y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate_synthetic(size, dim=6, noise=0.1):\n",
    "    X = np.random.randn(size, dim)\n",
    "    w = np.random.randn(dim + 1)\n",
    "    noise = noise * np.random.randn(size)\n",
    "    y = X.dot(w[1:]) + w[0] + noise\n",
    "    return X, y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Задание 1 (1 балл)\n",
    "Для начала нужно понять, какую метрику для ошибки будем использовать. В нашем случае нам подойдет стандартная метрика MSE. Ее и нужно реализовать."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def mse(y_true, y_predicted):\n",
    "    return np.mean((y_true - y_predicted)**2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Задание 2 (3 балла)\n",
    "Теперь реализуем линейную регрессию при помощи явного решения задачи минимизации. \n",
    "\n",
    "#### Методы\n",
    "`fit(X, y)` - решает задачу минимизации $\\arg\\min_{w, b}\\sum ((w\\cdot x + b) - y)^2$. \n",
    "\n",
    "`predict(X)` - строит предсказание `y` для объектов из `X`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy.linalg import inv\n",
    "\n",
    "class NormalLR:\n",
    "    def __init__(self):\n",
    "        pass\n",
    "    \n",
    "    def fit(self, X, y):\n",
    "        xx = np.c_[np.full(X.shape[0], 1), X]\n",
    "        self.w = np.dot(np.dot(inv(np.dot(xx.T, xx)), xx.T), y)\n",
    "    \n",
    "    def predict(self, X):\n",
    "        xx = np.c_[np.full(X.shape[0], 1), X]\n",
    "        return np.dot(xx, self.w)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-3.719664196232485 4.444989248515545\n"
     ]
    }
   ],
   "source": [
    "X, y = generate_synthetic(1024)\n",
    "print(np.min(X), np.max(X))\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.009849022845383117\n"
     ]
    }
   ],
   "source": [
    "regr = NormalLR()\n",
    "regr.fit(X_train, y_train)\n",
    "print(mse(y_test, regr.predict(X_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Задание 3 (2 балла)\n",
    "Теперь реализуем линейную регрессию с использованием градиентного спуска с larning rate `alpha` в течении `iterations` итераций.\n",
    "\n",
    "#### Методы\n",
    "`fit(X, y)` - приближает решение задачи минимизации $\\arg\\min_{w, b}\\sum ((w\\cdot x + b) - y)^2$ при помощи градиентного спуска. \n",
    "\n",
    "\n",
    "`predict(X)` - строит предсказание `y` для объектов из `X`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Задание 4 (2 балла)\n",
    "Добавьте в метод `fit` регуляризацию Лассо с коэффициентом `l`. Постройте график зависимости ошибки предсказания данных из синтетического набора данных от коэффициента регуляризации."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GradientLR:\n",
    "    def __init__(self, alpha, iterations=10000, l=0.):\n",
    "        self.alpha = alpha\n",
    "        self.iter = iterations\n",
    "        self.l = l\n",
    "    \n",
    "    def fit(self, X, y):\n",
    "        self.w = np.random.uniform(-0.1, 0.1, X.shape[1] + 1)\n",
    "        xx = np.c_[np.full(X.shape[0], 1), X]\n",
    "        for _ in range(self.iter):\n",
    "            y_pred = np.dot(xx, self.w)\n",
    "            self.w += (2 * self.alpha * (np.dot(xx.T, y-y_pred)) - self.l * np.sign(self.w)) / xx.shape[0] / xx.shape[1]\n",
    "\n",
    "    def predict(self, X):\n",
    "        xx = np.c_[np.full(X.shape[0], 1), X]\n",
    "        return np.dot(xx, self.w)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_plot(X_train, y_train, X_test, y_test, iterations=10000):\n",
    "    ll = np.arange(0, 0.2, 0.01)\n",
    "    ans = []\n",
    "    for l in ll:\n",
    "        regr = GradientLR(0.1, iterations=iterations, l=l)\n",
    "        regr.fit(X_train, y_train)\n",
    "        ans.append(mse(y_test, regr.predict(X_test)))\n",
    "    plt.plot(ll, ans)\n",
    "    plt.xlabel(\"Lasso l\")\n",
    "    plt.ylabel(\"MSE\")\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "X, y = generate_synthetic(1024)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.009234144251373496\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "regr = GradientLR(0.1, iterations=10000)\n",
    "regr.fit(X_train, y_train)\n",
    "print(mse(y_test, regr.predict(X_test)))\n",
    "build_plot(X_train, y_train, X_test, y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Задание 5 (2 балла)\n",
    "Протесируйте оба метода на данных cancer и spam, для градиентного спуска постройте график зависимости ошибки от коэффициента регуляризации. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "CANCER_DATA_PATH = \"hw9_data/cancer.csv\"\n",
    "SPAM_DATA_PATH = \"hw9_data/spam.csv\"\n",
    "def read_cancer_dataset(path_to_csv):\n",
    "    # Возвращает пару из X и y. X - массив векторов. y - соответствующие векторам метки\n",
    "    df = pd.read_csv(path_to_csv, header=0)\n",
    "    y = [1 if i == 'M' else 0 for i in df[\"label\"]]\n",
    "    return df.drop(\"label\", axis=1).to_numpy(), np.array(y)\n",
    "\n",
    "def read_spam_dataset(path_to_csv):\n",
    "    # Возвращает пару из X и y. X - массив векторов. y - соответствующие векторам метки\n",
    "    df = pd.read_csv(path_to_csv, header=0)\n",
    "    y = df[\"label\"]\n",
    "    return df.drop(\"label\", axis=1).to_numpy(), np.array(y.values)\n",
    "\n",
    "X_cancer, y_cancer = read_cancer_dataset(CANCER_DATA_PATH)\n",
    "X_cancer = StandardScaler().fit_transform(X_cancer)\n",
    "#print(np.min(X_cancer), np.max(X_cancer))\n",
    "X_spam, y_spam = read_spam_dataset(SPAM_DATA_PATH)\n",
    "X_spam = StandardScaler().fit_transform(X_spam)\n",
    "#print(np.min(X_spam), np.max(X_spam))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NormalLR 0.059868759834403394\n",
      "GradientLR 0.060755211937598586\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZUAAAEGCAYAAACtqQjWAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAAA8JElEQVR4nO3deXxV9Zn48c+TnUBCWAIJIexhC7tBcQEXtIgbrhWnHW07HYtVp/46m06Xmen87K/LtB3tWByrtTqjZRAXEBFE3LUoQUIWQiSsyc3KloSErPf5/XFP8BISst0tyfN+ve4r537P95zzPZdLnpzvKqqKMcYY4wthwS6AMcaY/sOCijHGGJ+xoGKMMcZnLKgYY4zxGQsqxhhjfCYi2AUIppEjR+qECROCXQxjjOlTdu7ceVRVE9vbN6CDyoQJE8jMzAx2MYwxpk8RkcMd7bPqL2OMMT5jQcUYY4zPWFAxxhjjMxZUjDHG+IwFFWOMMT7j16AiIteKSIGIFIrIw+3sFxF53NmfLSILvPYliMg6EdkrIvkicrGT/m9O3iwReUtExjjpE0TktJOeJSJP+vPejDHGnMtvQUVEwoEngOXATOAuEZnZJttyIM153Qus9tr3GLBZVacDc4F8J/2XqjpHVecBG4Efex2zX1XnOa9Vvr4nY4wx5+fPJ5ULgUJVPaCqjcAaYEWbPCuA59VjO5AgIskiEg8sAZ4BUNVGVT3pbFd7HT8YsLn7jTGmG5775BAf7Tvql3P7M6ikAEVe74udtK7kmQRUAs+KyC4ReVpEBrdmEpFHRaQI+BpnP6lMdPK/LyKL2yuUiNwrIpkikllZWdnjmzPGmL7o8yMn+MnGPazbWdR55h7wZ1CRdtLaPlV0lCcCWACsVtX5QC1wpk1GVX+gqqnAC8ADTnIpMM7J/33gReeJ5+yTqz6lqhmqmpGY2O4sA8YY0y+damjmoTVZJMXH8JObZ/nlGv4MKsVAqtf7sUBJF/MUA8Wq+qmTvg5PkGnrReA2AFVtUNVjzvZOYD8wtZf3YIwx/cY/r8+j+EQdj62cR3xMpF+u4c+gsgNIE5GJIhIFrAQ2tMmzAbjb6QW2CKhS1VJVLQOKRGSak28psAdARNK8jr8J2OukJzqdAxCRSXga/w/46d6MMaZPeX13CS9/XswDV6WRMWG4367jtwklVbVZRB4AtgDhwB9UNU9EVjn7nwQ2AdcBhUAd8E2vUzwIvOAEpANe+37mBBs3cBho7eW1BPiJiDQDLcAqVT3ur/szxpi+wnXyNP/0ag7zxyXwN1dN8eu1RHXgdp7KyMhQm6XYGNOftbiVu36/nTxXFZu+t5jxIwZ3flAnRGSnqma0t29AT31vjDH93ZPv7+ezg8f51R1zfRJQOmPTtBhjTD+VVXSS32z9ghvmJHPrgrYjOvzDgooxxvRDtQ3NPLRmF6PjY3j0ltmItDeCw/es+ssYY/qhf309j8PH61jz14sYOsg/3YfbY08qxhjTz2zKKWVtZjHfvWIyF00aEdBrW1Axxph+pOTkaR55JYe5Y4fy0NWBH/9tQcUYY/qJFrfy/bVZNLW4eWzlfCLDA/8r3tpUjDGmn/j9hwfYfuA4v7h9DhNG+r/7cHvsScUYY/qBnOIq/n1LAdfNTuKOC8YGrRwWVIwxpo+ra2zme2t2kRgXzU8D2H24PVb9ZYwxfdy/bdzDwWO1vPDti0iIjQpqWexJxRhj+rDNuWX86bMivrNkMpdMHhns4lhQMcaYvqqsqp6HX8lmdspQvn9NaCwfZUHFGGP6ILdb+duXsmhocvMfK+cRFREav85DoxTGGGO65ZmPDvJx4TF+fONMJicOCXZxzrCgYowxfUyuq4pfbNnLsvTRrFyY2vkBAWRBxRhj+pgfvJbL8MFR/OzWOUHtPtweCyrGGNOH1DY0k118kpULxzFscHC7D7fHgooxxvQhe0qrUYXZKUODXZR2+TWoiMi1IlIgIoUi8nA7+0VEHnf2Z4vIAq99CSKyTkT2iki+iFzspP+bkzdLRN4SkTFexzzinKtARJb5896MMSYYcoqrAJg9doAFFREJB54AlgMzgbtEZGabbMuBNOd1L7Daa99jwGZVnQ7MBfKd9F+q6hxVnQdsBH7sXG8msBJIB64FfueUwRhj+o3ckioS46IZHR8T7KK0y59PKhcChap6QFUbgTXAijZ5VgDPq8d2IEFEkkUkHlgCPAOgqo2qetLZrvY6fjCgXudao6oNqnoQKHTKYIwx/Uauqypkq77Av0ElBSjyel/spHUlzySgEnhWRHaJyNMicmYeZxF5VESKgK/hPKl08XqIyL0ikikimZWVlT27M2OMCYK6xmYKK04xa4AGlfb6uWkX80QAC4DVqjofqAXOtMmo6g9UNRV4AXigG9dDVZ9S1QxVzUhMTOz8LowxJkTkl1bjVpg1Jj7YRemQP4NKMeA9KmcsUNLFPMVAsap+6qSvwxNk2noRuK0b1zPGmD4r1Bvpwb9BZQeQJiITRSQKTyP6hjZ5NgB3O73AFgFVqlqqqmVAkYhMc/ItBfYAiEia1/E3AXu9zrVSRKJFZCKexv/P/HJnxhgTBDmuakYOiSIpRBvpwY/rqahqs4g8AGwBwoE/qGqeiKxy9j8JbAKuw9OoXgd80+sUDwIvOAHpgNe+nznBxg0cBlrPlycia/EEn2bgflVt8df9GWNMoOW6qpiVMjTkRtF78+siXaq6CU/g8E570mtbgfs7ODYLyGgn/bZzc5/Z9yjwaA+La4wxIet0Ywv7Kmr4SvroYBflvGxEvTHG9AF7WhvpQ7jnF1hQMcaYPiGvxGmkt6BijDGmt3KKqxgxOIrkoaHbSA8WVIwxpk/IcVWRHuKN9GBBxRhjQl59Uwv7Kk4xOyV0Bz22sqBijDEhLr+0mha3hnx7ClhQMcaYkJfr8jTSh3rPL7CgYowxIS/HVcWw2EhSEgYFuyidsqBijDEhLsdVHfIj6VtZUDHGmBBW39TCvvKaPtGeAhZUjDEmpBWU1dDcRxrpwYKKMcaEtJw+1EgPFlSMMSak5bqqGDookrHDQr+RHiyoGGNMSMtx1qTvC430YEHFGGNCVkNzC1+U1/SZqi+woGKMMSGroKyGppa+00gPFlSMMSZktTbSW1AxxhjTa62N9KnD+0YjPfg5qIjItSJSICKFIvJwO/tFRB539meLyAKvfQkisk5E9opIvohc7KT/0knLFpFXRSTBSZ8gIqdFJMt5Pdn2esYY05fkuqqZlRLfZxrpwY9BRUTCgSeA5cBM4C4Rmdkm23IgzXndC6z22vcYsFlVpwNzgXwnfSswS1XnAF8Aj3gds19V5zmvVb6+J2OMCZTGZjcFZX2rkR78+6RyIVCoqgdUtRFYA6xok2cF8Lx6bAcSRCRZROKBJcAzAKraqKonne23VLXZOX47MNaP92CMMUHxRXkNjS3uPtWeAv4NKilAkdf7YietK3kmAZXAsyKyS0SeFpHB7VzjW8CbXu8nOvnfF5HFvb4DY4wJkjMj6cdYUGnVXiWgdjFPBLAAWK2q84Fa4Kw2GRH5AdAMvOAklQLjnPzfB150nnhoc9y9IpIpIpmVlZXduR9jjAmYHFcVcTERjB8RG+yidIs/g0oxkOr1fixQ0sU8xUCxqn7qpK/DE2QAEJF7gBuAr6mqAqhqg6oec7Z3AvuBqW0LpapPqWqGqmYkJib24vaMMcZ/cl1VzBrTd0bSt/JnUNkBpInIRBGJAlYCG9rk2QDc7fQCWwRUqWqpqpYBRSIyzcm3FNgDnh5lwD8CN6lqXeuJRCTR6RyAiEzC0/h/wI/3Z4wxftHY7GZvaQ2zx/atqi/wVDP5hao2i8gDwBYgHPiDquaJyCpn/5PAJuA6oBCoA77pdYoHgRecgHTAa99/AtHAVieCb3d6ei0BfiIizUALsEpVj/vr/owxxl9aG+n7Ws8v8GNQAVDVTXgCh3fak17bCtzfwbFZQEY76VM6yP8y8HIvimuMMSEhr6TvjaRvZSPqjTEmxOS4qoiLjmD88L7VSA8WVIwxJuTkuKpJT4knLKxvNdKDBRVjjAkpTS1u8kur+9z4lFYWVIwxJoTsKz9FY7O7T/b8AgsqxhgTUnL72Jr0bVlQMcaYEJLjqmJIdAQTR7Q3M1Xos6BijDEhJMdVxcwxfbORHiyoGGNMyGh2Gun74viUVhZUjDEmRBRWnqKhue9Nd+/NgooxxoSInOK+3UgPFlSMMSZk5LqqGBwVzqSRfbORHiyoGGNMyOjrjfRgQcUYY0JCc4ubPaXVfbrqC/w8S7ExxoSin2/ey+FjtYyOjyEpPoakoTFnbcdEhge8TPsra6lv6tuN9GBBxRgzwBw6Wsvq9/Yzckg0pxubqW1sOSfP0EGRJMXHMHpoDEnx0V7bnuCTPDSGEUOifVqu1jXpLagYY0wfsmF3CSLw+oOXkjx0EDX1TZRX11NaVU9ZVT3l1fWUVddTVtVAeXU9+aXVHD3VgGfh8i99+7KJ/PCGmT4rV66ritiocCYlDvHZOYPBgooxZsBQVV7LcnHhhOEkDx0EQFxMJHExkUwZFdfhcU0tbiprGiirrqe8qp6NOaU88/FBbpg7hnmpCT4pW66ripnJ8YT34UZ6sIZ6Y0wQ5JVUUVXXFITrVnOgspYV81K6dVxkeBhjEgaxYNwwls9O5me3zmbkkGh+vD6XFrd2foJOtLiVvJK+30gPFlSMMQF2qqGZW3/3Cf/6el7Ar70+y0VkuLB8VlKvzhMXE8kPr59BdnEVa3Yc6XW5DlSe4nRTS59vTwE/BxURuVZECkSkUEQebme/iMjjzv5sEVngtS9BRNaJyF4RyReRi530Xzpp2SLyqogkeB3ziHOuAhFZ5s97M8b0zEf7KmlodvNGTilVpwP3tNLiVjbsLuHyqYkMGxzV6/PdNHcMiyYN5xebCzhe29irc51ppO+ja6h481tQEZFw4AlgOTATuEtE2rZqLQfSnNe9wGqvfY8Bm1V1OjAXyHfStwKzVHUO8AXwiHO9mcBKIB24FvidUwZjTAh5O7+CqPAwGprdbMhyBey6nx08Tnl1Q7ervjoiIvxkxSxqG5r5xea9vTpXjquKmMiwPj2SvpU/n1QuBApV9YCqNgJrgBVt8qwAnleP7UCCiCSLSDywBHgGQFUbVfWks/2WqjY7x28Hxnqda42qNqjqQaDQKYMxJkS43cq7eyu4dlYSM5PjWbOjKGDX3rDbRWxUOFfPGO2zc04dHce3LpvImh1FfH7kRI/P09pIHxHe91sk/HkHKYD3N6bYSetKnklAJfCsiOwSkadFpL0Q/i3gzW5cDxG5V0QyRSSzsrKyO/djjOml3cUnOVbbyNIZo1h5YSp5JdVnVjr0p4bmFjbllLEsPYlBUb6twPibpWmMju95o31rI31/aE8B/waV9vrFtf3EO8oTASwAVqvqfKAWOKtNRkR+ADQDL3TjeqjqU6qaoaoZiYmJ578DY4xPbcuvIDxMuHxqIivmphAdEcb/BuBp5YMvjlJ1uomb5o3x+bmHREfww+tnkuuq5sVPD3f7+INHT1HX2NIven6Bf4NKMZDq9X4sUNLFPMVAsap+6qSvwxNkABCRe4AbgK+pnhmS1JXrGWOCaNveCi4YP4yE2CiGxkZy3exkXstycbqdUe2+tD7LxfDBUVw2ZaRfzn/DnGQumTyCX24p4Oiphm4d258a6cG/QWUHkCYiE0UkCk8j+oY2eTYAdzu9wBYBVapaqqplQJGITHPyLQX2gKdHGfCPwE2qWtfmXCtFJFpEJuJp/P/Mb3dnjOkW18nT5JdWs3T6qDNpX81Ipaa+mTdzS/123VMNzbydX871s5OJ9FObhafRPp3TTS38/M3uNdrnuqqJiQxjSh8fSd/Kb0HFaUx/ANiCp+fWWlXNE5FVIrLKybYJOICnUf33wHe9TvEg8IKIZAPzgJ866f8JxAFbRSRLRJ50rpcHrMUTfDYD96uqf//8McZ02Tt7KwBY6tVQvmjScCaMiPVrFdhbeWXUN7lZ4YeqL29TRsXxV5dN4qWdxew8fLzLx+W4qpjRTxrpwc/TtKjqJjyBwzvtSa9tBe7v4NgsIKOd9Cnnud6jwKM9LK4xxo/eyS9n/IhYJid+2edGRPjqwlR+sbmAA5Wn/DLv1fqsElKc0fD+9uBVU1if5eJHr+Wx4YFLOw0Ubreyp6SaWxf4pptzKOgfodEYE9LqGpv5eP8xlk4fjcjZfWpuXzCW8DBhbWaxz6979FQDHxUe5aZ5YwKy8NXg6Ah+dMNM9pRW88KnnY+0P3isllMNzcwa0z/aU8CCijEmAD4uPEZjs5ulM0ads29UfAxXThvFup3FNLW4fXrdTTmltLiVm3004LErls9KYnHaSP79rQIqa87faN/anbq/9PyCToKKiHzda/vSNvse8FehjDH9y7b8cuKiI1g4YXi7+1cuTOXoqQbeddpdfGV9VgnTk+KYltTxDMS+JiL8y03p1De18P/ezD9v3pziKqIiwkgb3T8a6aHzJ5Xve23/ts2+b/m4LMaYfsjtVt7ZW8GSqYlERbT/K+eKaYmMiov2aYN90fE6dh4+4ZexKZ2ZnDiEv148iVc+d/HZwY4b7Vsb6f3VKy0YOrsT6WC7vffGGHOOvJJqKmoauGr6uVVfrSLCw7j9grG8W1BBWVW9T667YbdnmNqNcwIfVAAeuGoKKQmD+PH6XJrbqdZznxlJHx+E0vlPZ0FFO9hu770xxpzj7fxyRODK8wQV8IxZcSus2+mbp5UNWSVkjB9G6vBYn5yvu2KjPI32e8tqeP7P5460P3y8jlMNzf1mepZWnQWV6c4U8zle263vp3VyrDHG8M7eChaMG8bwTqabnzByMBdPGsHazGLcvVz4am9ZNQXlNX4fm9KZZemjuXxqIr/Z+gUV1Wc/geX0w0Z66DyozABuxDMlSut263vfLc5sjOmXyqvryXFVnbfqy9udC1M5cryO7QeO9eq667NKCA8Trpud3Kvz9FZro31Ds5v/12akfa7L00g/dXTgOhEEwnmDiqoe9n4Bp/DMwTXSeW+MMR1qHUXf1enmr52VRHxMRK+mxHe7lQ1ZJSxOG8mIIdE9Po+vTBw5mO9cPolXd7nOCpY5xVXMSIrrV4300HmX4o0iMsvZTgZy8fT6+m8Recj/xTPG9GXb8itISRjE1C52mY2JDOeW+SlszivjZF3PVlPceeQErpOng1715e27V3zZaN/U4kZVyS2pIr2fVX1B59VfE1U119n+JrBVVW8ELsK6FBtjzqO+qYWPCiu5esaoc0bRn8+dC8fR2OzmtV09WxVyfZaLmMgwvjKzd+vQ+9KgqHD++caZfFF+iuc+OcThY3XU1Pe/RnroPKh4LyC9FGceL1WtAXw79NUY06/8ef8x6pvcXNXNlRZnjolndspQ1uwo4suVLbqmqcXNG9mlXDMzicHRfp3asNuumTmaq6aP4jdbv+Dt/HKAARlUikTkQRG5BU9bymYAERkERPq7cMYY32tsdnPjbz9izWedz03VG9v2lhMbFc5FE9sfRX8+dy5MZW9ZDdnF3VsV8qN9RzlR18SKuaFT9dVKRPjnG2fS5FZ+sbmAqPD+10gPnQeVvwLSgW8Ad7auEw8sAp71X7GMMf7y/heV5Liq+MWWAmobmv1yDVXlnfwKFqeNJCay+8v33jRvDDGRYfxvZvca7NdnuRg6KJIlU0NzVdfxIwZz3+WTaWxxMy0prsMZBvqyznp/VajqKlVdoapveaW/q6r/7v/iGWN87bVdLgZFhnO8tpE/fnLIL9fIL62hpKqepdO7V/XVKj4mkutnj2FDVgl1jV0LfHWNzby1p5zrZieH9C/r+66YTNqoISxO888qlMF23kpHEWm7UuNZVPUm3xbHGONP1fVNbM0v566FqRSdOM1THxzgLy8eT3yMb2uztzltBp2Noj+fOxem8vLnxbyRXcodGamd5n87v4K6xpaQ6vXVnpjIcLY8tCQgU/EHQ2ctWRcDRcCfgE+x+b6M6dPezCmlsdnNLQvGEhEm3PDbj/jDRwd56OqpPr3Otr0VzE1NIDGu5+NEFk4YxqSRg1mbWdSloLIhy0VSfAwXdjATcijprwEFOm9TSQL+CZgFPAZcAxxV1fdV9X1/F84Y41uv7nIxceRg5o4dyqyUoSxLH80zHx7s8ZiQ9lTWNLC7+ORZa9H3hIhw58JUdhw6QWHFqfPmPVHbyHsFlQFbjMt0rLM2lRZV3ayq9+BpnC8E3hORBwNSOmPaUVXXxFW/eu/MLLSma1wnT7P9wHFunpdyZtzI/7lmKqcam/n9hwd8dp13CypQpd0FubrrVueJam0nDfZv5pbR7FZuCsFeXwNNp61ZIhItIrcC/4NnPfnHgVe6cnIRuVZECkSkUEQebme/iMjjzv5sEVngtS9BRNaJyF4RyReRi530O0QkT0TcIpLhlX+CiJwWkSzn9WRXymj6nvW7XRyorOUHr+TgOnk62MXpM9ZneQYT3jL/y1UQpyfFc/3sZJ79+BDHTp1/lcKueie/guShMcxM7v2U7olx0SydMYqXdxbT2Nzx0LjXslxMThxM+pj+NY18X9TZNC3PAZ/gGaPyr6q6UFX/TVU7HeoqIuHAE8ByPJNP3iUibSehXA6kOa97gdVe+x4DNqvqdGAu0LqEWi5wK/BBO5fdr6rznNeqzspo+qa1mUVMGBGLW5W/f2l3r2e0HQhUlVc/d3HB+GGMG3H2VPAPXT2V+qYWnnx/f6+v09Dcwof7KrlqevdG0Z/PyoXjOFbbyDt7y9vdX3LyNJ8dPPsJzARPZ08qfwlMBb4HfCIi1c6rRkSqOzn2QqBQVQ+oaiOwBljRJs8K4Hn12A4kiEiyiMQDS4BnAFS1sXWMjKrmq2pBd27S9B97SqrJdVXzzUsn8qMbZvLJ/mM89+dDwS5WyMsrqWZfxSlunn/uWu1TRg3h5nkpPP/nw+dMz95dnx44Tm1ji0+qvlotmZpIUnxMh5NMvu5UgwZjhUdzrs7aVMJUNc55xXu94lS1s+fMFDw9x1oVO2ldyTMJqASeFZFdIvK0iAzuwv1MdPK/LyKL28sgIveKSKaIZFZWVnbhlCaUvLSziKjwMFbMG8OdC1O5avoofvbm3k4bcge613a5iAwXbuhgKvi/WZpGs1v53Xu9e1p5Z28FMZFhXDLZd2MwwsOEOzLG8v4XlZS0U925PquEeakJjB/RlV8Rxt/8OUKovefQtvUUHeWJwFPltlpV5wO1wDltMm2UAuOc/N8HXnSeeM4+uepTqpqhqhmJiT0bdVtRU8/q9/b7rA7adE3rJIPXpI8mITYKEeFnt80mNiqc76/NoqmdJVsNtLiV9btLuGLaKIZ1sFDWhJGDuX3BWF789Ei7v7i7QlV5O7+cy6b0bBT9+Xw1IxVVeCmz+Kz0feU17CmtDvmxKQOJP4NKMeDduXws0La7Tkd5ioFiVf3USV+HJ8h0SFUbVPWYs70T2I+n6s7nKmsa+PnmvWcmhTOBsS2/nBN1TdxxwdgzaaPiYnj0ltlkF1fxu3d73ybQH31ceJTKmoazGujb8+DSKSjKf75b2KPr7Ks4RfGJ01zVw1H055M6PJbLpoxkbWbRWW1oG3aXECZw/ZzgLsZlvuTPoLIDSBORiSISBawE2o7Q3wDc7fQCWwRUqWqpqpbhmcyydcnipcCe811MRBKdzgGIyCQ8jf++6yfpZWZyPGOHDWJLngWVQFqbWURSfAyL085+wrxudjI3zxvDb9/ZR3bxyeAULoS9tstFXExEp6svjh0Wy50LU1m7o4ii43Xdvk7rH1ldXeWxu766MBXXydN8vP8o4HkyWp9VwqVTRjIqLsYv1zTd57egoqrNwAPAFjw9t9aqap6IrBKR1p5Zm/D84i8Efg981+sUDwIviEg2MA/4KYCI3CIixXhG+78hIluc/EuAbBHZjefJZpWqHvfHvYkIy9KT+GjfUU75aUI+c7ayqnre/6KS2y8YS3g7g9v+9aZZjBwSzffX7qa+qSUIJQxNdY3NbM4r4/rZyV2qknrgyjTCwoTHt+3r9rXeya9gVko8SUP98wv+KzNHkxAbeabBPqvoJEeO19nYlBDj11nXVHWTqk5V1cmq+qiT9qSqPulsq6re7+yfraqZXsdmOW0fc1T1ZlU94aS/qqpjVTVaVUer6jIn/WVVTVfVuaq6QFVf9+e9LUtPorHFzXsFFf68jHG8sqsYt8LtXlVf3obGRvLLO+ZQWHGKX26xzoGt3sorp66xpdOqr1ZJQ2P4+kXjeWWXi4NHa7t8neO1jXx+5IRfqr5ata4KuTWvnOO1jazPKiEqIoxls0JnMS7j56DSn10wfhgjBkdZFVgAqCovZRZz4cThTBjZcQ+fxWmJ/OWi8Tzz0UE+capIBrpXd7lISRjEwm7Mh3XfFZOJCg/jsbe/6PIx7xVU4Fa42oddidtz58JUGlvcrNtZxMbsUpZOH+XzyTBN71hQ6aHwMOGamaN5d28FDc1W3eJPOw+f4ODR2rMa6DvyyHXTmTAilr9/KZua+qZO8/dnlTUNfLivkhXdnA8rMS6auy8Zz/rdJewrr+nSMdv2VpAYF82sMf5dyXB6UjxzUxP4j7f3cfRUAyvmde0JzASOBZVeWJaexKmGZj7ZfyzYRenX1mYWMTgqnOs6GGPhLTYqgl99dR6lVaf5yevn7dvR723YXYJb6XLVl7fvLJlMbGQ4//F2520rjc1uPiioZOn0UQGZzHHlwlTqGluIi4ngimmhuRjXQGZBpRcumTKCIdERvJVXFuyi9Fu1Dc1szC7lhjljurzm+AXjh3HfFZN5aWfxgP63eW2Xi1kp8aT1YMna4YOj+NZlE3kjp5Q9JeefPCPz0HFqGpr91uurrRvnjiEuOoIb5nSt84EJLAsqvRAdEc4V0xLZuqecFpt/yi825ZRS19jCHRmdV315+97SqcxIjueRV3I4OgAHqRZW1JDjquLmXlQPffuyScTFRPCbTtpWtu2tICoijMsCtJLhkOgINn1vMT+8vu1UgiYUWFDppa+kJ3H0lKfni/G9lzKLmTRyMBeMH9at46IiwvjNnXOpqW/mB6/moDqwgv6ru1yESe/mwxoaG8lfL57E1j3lHY7/UVW25ZdzyeQRxEZ17UnSF1KHx3b5ydUElgWVXrpyWiJR4WFsyR241Sz+cvBoLZ8dOs7tGWN7NPvs9KR4/vYrU9mSV84rn3c6sXa/4XYrr+0q4bK0xF4PCvzmpRNIiI3k11vbf1o5cLSWQ8fqer0gl+k/LKj0UlxMJJdMGcGWPWUD7q9hf1u3s4gwgdsWdK/qy9u3F09i4YRh/MuGvAGz9sqOQ8dxnTzNLfN7PygwLiaS7yyZzHsFlew8fO7T+Dv5nnFaV83w3/gU07dYUPGBZelJFB0/TX5p17pfms61uJWXd7q4YtooRsf3/K/t8DDhV3fMo2UArb3yWpaL2KhwlqX7ZlDgPZeMZ+SQKH699dxBpW/nlzM9KY6UhEE+uZbp+yyo+MDVM0YjAlsGcE8jX/twXyVl1fVdGpvSmXEjYgfM2iv1TS1szC5lWXqSz9o4YqMiWHX5ZD4uPMafvbrPV9U1kXn4hE/XTjF9nwUVH0iMiyZj/DALKj70UmYxwwdHsdRH1SorF6Zy5bTEfr/2yrt7K6ipb253Ma7e+Pqi8YyOj+bXWwvOVPO+90UFLW712b+R6R8sqPjIsvQk9pbVcORY92d3NWc7UdvI1j3lrJg3hqgI33xFRYSf3zaHQVHh/O3aLJr76dorr+5ykRgXzaWTR/j0vDGR4dx/5RR2HDrBh/s8U+C8s7eCEYOjmDs2wafXMn2bBRUfaa2/tqeV3luf5aKxxc0dF6R2nrkbRsXH8OjNs9ldXNXrFQ5D0cm6Rt4tqOCmuWOICPf9f+07F6YyZmgMv9r6Bc0tbt4rqOTK6aPanTXaDFwWVHwkdXgsM5LjLaj4wNrMYmanDGXmmM5WrO6+6+cks2LeGB7fto9cV5XPzx9MG7NLaWrRHk3L0hXREeE8uDSN3UUn+eVbBVSdbrKuxOYcFlR8aFn6aHYeOUFlzcAbwe0rua4q9pRWd3sEfXf85KZZDI6OYHU/e1p5bZeLtFFDSPdDMG51+wVjGTc8lv96/wCR4RKwUfSm77Cg4kPL0pNQha17bDr8nlq3s5ioiDC/Lrw0NDaSlQtT2ZxX1uP12EPNkWN1ZB4+wc3zU3o0ULSrIsPD+JulaQAsmjSCOJt23rRhQcWHpifFMW54rFWB9VBDcwuvZbmcFf6i/Hqtry8aj6ryP9sP+/U6gfJalmfGAF/3+mrPzfPGcMOcZL5xyQS/X8v0PRZUfMizzPBoPtl/lOoBvpZHT7y9p4KTdU18NcO3DfTtSR0ey9UzRvOnz470+eWHVZXXdrm4aOLwgAxCjAgP4z//YoF1JTbtsqDiY8vSk2hqUd7da8sMd9fazCLGDI3h0imBqaf/xqUTOFHXxIbdJQG5nr9kF1dx4Git3xrojekOvwYVEblWRApEpFBEHm5nv4jI487+bBFZ4LUvQUTWicheEckXkYud9DtEJE9E3CKS0eZ8jzjnKhCRZf68t44sGDeMkUOiecuWGe6W0qrTfLivktsuGBuwLqoXTxrBtNFxPPfJoT49b9uru1xERYSxvAuLmBnjb34LKiISDjwBLAdmAneJSNsFEJYDac7rXmC1177HgM2qOh2YC+Q76bnArcAHba43E1gJpAPXAr9zyhBQYc4yw+8VVPT5apVAeuVzF2719C4KFBHhnksmkFdSTWY7kyX2BU0tbl7fXcLVM0YxdJA1mpvg8+eTyoVAoaoeUNVGYA2wok2eFcDz6rEdSBCRZBGJB5YAzwCoaqOqnnS281X13JntPOdao6oNqnoQKHTKEHDL0kdT29jCx4VHg3H5PkdVeSmziIsmDmf8iMEBvfbN88cwdFAkf/z4UECv6ysf7TvKsdrGXi3GZYwv+TOopABFXu+LnbSu5JkEVALPisguEXlaRDr7bdOV6yEi94pIpohkVlZWdu1OuumSySOJi46wXmBdtOPQCQ4dqwtIA31bsVER3Ol0Ly6t6nvdi1/Z5SIhNpIrptkgRBMa/BlU2qsYb1tx3VGeCGABsFpV5wO1wDltMj24Hqr6lKpmqGpGYmJiJ6fsmaiIMK6cPoq38yv67RxTvrQ2s4gh0REsn+2bqdq76y/7aPfimvom3sor44Y5yT6bI82Y3vLnN7EY8P7TcyzQtptNR3mKgWJV/dRJX4cnyPT2egGzLD2J47WNfbauPlBONTSzKaeUG+YkB3Q5Wm9fdi8u6lPtYJtzy2hodnPL/MC1QxnTGX8GlR1AmohMFJEoPI3oG9rk2QDc7fQCWwRUqWqpqpYBRSIyzcm3FNjTyfU2ACtFJFpEJuJp/P/MZ3fTTVdMSyQqIsx6gXViU3YpdY0t3BGEqi9v37hkAsdrG3m9D3Uvfi3LxbjhsSwYlxDsohhzht+Ciqo2Aw8AW/D03FqrqnkiskpEVjnZNgEH8DSq/x74rtcpHgReEJFsYB7wUwARuUVEioGLgTdEZItzvTxgLZ7gsxm4X1WD9mfn4OgIFk8ZyZY8W2b4fNZmFjEpcXDQfzFePHkEU0cP4Y99pHtxWVU9n+w/5vdpWYzpLr/WN6jqJjyBwzvtSa9tBe7v4NgsIKOd9FeBVzs45lHg0Z6X2LeWpSexbW8FeSXVzEoZGuzihJwDlafIPHyCh5dPD/ovRhHhG5dM5J9ezSHz8AkWThge1PJ0Zn2WC1VswKMJOda650dLZ4wiTOAt6wXWrpd2FhMeJtwaIr8Yb54/hviYCP74yaFgF+W89pXX8L+ZRcxLTWDiyMB2wTamMxZU/GjEkGgWThjOFmtXOUdzi5tXPi/miqmJjIqPCXZxAE/34pUXjmNzbuh1Lz56qoE/fHSQG3/7Edf85gMOH6vjrxdPCnaxjDmHBRU/W5aeREF5DYeO1ga7KCHlw31HKa9uCHoDfVt/uWg8blVe2H4k2EWhvqmFjdklfOuPO7jop9v4ycY9KMqPbpjJ9keWcv0cm5bFhJ7g9OEcQL6SPpqfbNzDlrwyvnP55GAXJ2SszSxi+OAorgqxlQNbuxe/+NkRHrhqCjGRgZ3px+1Wdhw6zqu7XLyRXUpNQzNJ8TH89eJJ3Loghamj4wJaHmO6y4KKn40dFsuslPg+E1QamluoOt1EVV0TJ71+nqxrpPq0k+a8mlrcNLcoLW6lRT0/m1sUtyrNbue9243bDc1ut/Pek15T38xfXTYxJAftffOSCWzdU87ru0sC9iR1oPIUr+5y8crnLlwnTxMbFc61s5K4bcFYFk0aYevAmz7DgkoALJuZxK+2fkFFdX1ItB80Nrv5+ea9FJ+o42Tdl0HiZF0Tp88z+E8E4mMiSYiNJD4mkuiIMMLChKiIMMLDhIgwIdx5RYSFndluuy88TIiOCOdbl04I3E13g3f34tsvGOu3nmnHaxvZmF3CK5+7yCo6SZjApVNG8nfLprIsPSlog0GN6Q371gbAslmeoPLWnnK+vmh8sIvDO3vLeeajg0xKHMzIwdGkDo9l9qBIhg7yBIyhsVGe7db3gyJJGBRFXEwEYQPgL+bW2Yt/8GouOw+fIMPH3YtrG5p55JUc3swtpalFmZ4Uxz9dN50V81IYHQJ/dBjTGxZUAiBt1BAmjhzMlryykAgqr2eXMmJwFG89tISI8NCrfgoFt8xP4edv7uXZTw75NKicbmzhr57bwY5DJ/jGJRO4bcFYZo6J99n5jQk2+40SACLCV9JH8+f9x6g6HdxlhmsbmtmWX851s5MtoJzHmdmLfdi9uL6phXv/O5NPDx7n11+dy49umGkBxfQ79lslQJalJ9HsDv4yw2/nl1Pf5ObGuWOCWo6+4O6LJ/ise3FDcwv3/c9OPio8yi9vn8sKW//E9FMWVAJk3tgERsVFB32NlY3ZpSTFx5AxflhQy9EXeHcv7s3sxU0tbh54cRfvFlTy01tmB3R1S2MCzYJKgISFearA3iuoDNr06lWnm3i/oJLr5yQPiAZ3X2idvXhjdmmPjm9ucfO9NbvYuqecn6xI564Lx/m4hMaEFgsqAbQsPYnTTS18uC84ywy/lVdGY4tVfXXHJZNHkDZqCH/85GC3Zy9ucSt/+9JuNuWU8cPrZ3D3xRP8U0hjQogFlQC6aOII4mKCt8zwxuxSUocPYu5YmzG5q0SEb1w6gVxXNZ8f6fqCa2638o8vZ7M+q4S/XzaNb9s8XWaAsKASQFERYSydPopt+eUBX2b4eG0jHxUe5YY5Y4I+zXxfc8v8FOJjInj240Ndyq+q/HB9Lut2FvO9pWncf+UU/xbQmBBiQSXAlqUncaKuic8OHQ/odd/MLaXFrdw4x6q+uqu1e/GbuWWUVdWfN6+q8q+v7+HFT49w3xWTeejqtACV0pjQYEElwC6flkh0EJYZ3ri7lEmJg5mRbBMS9sSZ7sWfHu4wj6ry0035/PGTQ3z7son8w7Jp9lRoBhwLKgEWGxXB4rRE3grgMsMV1fVsP3iMG63qq8dSh8eydPpoXvy0/e7Fqsq/v1XA7z88yD0Xj+cH18+wz9oMSBZUgmBZ+mhKqurZXVwVkOu9kVOKKtw419bf6I1vXjqBYx10L358WyFPvLufuy5M5Z9vTLeAYgYsvwYVEblWRApEpFBEHm5nv4jI487+bBFZ4LUvQUTWicheEckXkYud9OEislVE9jk/hznpE0TktIhkOa8n/XlvvfGV9CQGR4Xz7McHA3K9jdmlTE+KY8ooq/rqjdbuxc99cuisp8zV7+3nN29/wW0LxvLozbNtDJAZ0PwWVEQkHHgCWA7MBO4SkZltsi0H0pzXvcBqr32PAZtVdTowF8h30h8GtqlqGrDNed9qv6rOc16rfH1PvjJ0UCR/cdE4Xt9dwpFjdX69luvkaXYePmFjU3ygdfbiHFfVme7FT394gJ9v3stNc8fwi9vnWEAxA54/n1QuBApV9YCqNgJrgBVt8qwAnleP7UCCiCSLSDywBHgGQFUbVfWk1zHPOdvPATf78R785tuLJxERFsZ/fbDfr9d5I7sEwHp9+citC1KIi4ngj58c5vk/H+L/vpHP8llJ/Pqrc20hLWPwb1BJAYq83hc7aV3JMwmoBJ4VkV0i8rSIDHbyjFbVUgDnp/d6tBOd/O+LyOL2CiUi94pIpohkVlZW9vjmemt0fAy3XZDCSzuLqag5fzfV3nh9dylzxw5l3IhYv11jIImNimDlwlTeyC7hx+vzuHrGKB5bOd9mfDbG4c//Ce392da2u1NHeSKABcBqVZ0P1HJ2NVd7SoFxTv7vAy86Tzxnn1z1KVXNUNWMxMTEzu7Br76zZDLNLW7+8NEhv5z/0NFaclxVVvXlY3dfPIGIsDAun5rIE19bEJJLIhsTLP7831AMeC/wPRYo6WKeYqBYVT910tfhCTIA5SKSDOD8rABQ1QZVPeZs7wT2A1N9djd+MGHkYJbPTuZ/th/2yzorG52qr+vnWK8vX0odHsv7/3AFz9yTQXREeLCLY0xI8WdQ2QGkichEEYkCVgIb2uTZANzt9AJbBFSpaqmqlgFFIjLNybcU2ON1zD3O9j3AegARSXQ6ByAik/A0/h/w0735zH2XT+ZUQzP/s73jQXU99fruUhZOGEby0EE+P/dAlzx0kFV5GdMOv/2vUNVm4AFgC56eW2tVNU9EVolIa8+sTXh+8RcCvwe+63WKB4EXRCQbmAf81En/GXCNiOwDrnHeg6dhP1tEduN5slmlqoGdC6UHZqUM5fKpifzho4OcbvTdlPhflNdQUF5jVV/GmIDy6xr1qroJT+DwTnvSa1uB+zs4NgvIaCf9GJ4nl7bpLwMv967EwfHdKyZz51PbeWlnkc+mR9+4u4QwgeWzrOrLGBM49vweAi6cOJwF4xL4r/cP0OSD2YtVldezS7l48ggS46J9UEJjjOkaCyohQET47hVTcJ08zeu72/Zl6L68kmoOHq21sSnGmICzoBIirpo+immj41j93n7c7t5NNPl6dgkRYcK1s5J8VDpjjOkaCyohIixMuO+KyeyrOMW2vRU9Po+qsnF3KYvTRpIQG+XDEhpjTOcsqISQG+YkM3bYIH73XmGPp8X//MhJXCdPW68vY0xQWFAJIRHhYXxnySR2HTnJ9gM96w29MbuEqIgwrpk52selM8aYzllQCTF3ZKQyckgUv3uvsNvHtriVN7JLuXJaInExkX4onTHGnJ8FlRATExnOty6byIf7jpLTzUW8Pjt4nIqaBqv6MsYEjQWVEPT1ReOJi47gyfe7Ny3+xuwSYqPCuWr6qM4zG2OMH1hQCUHxMZF8/eLxbMot5UDlqS4d09Ti5s3cMpbOGE1slF8nSjDGmA5ZUAlR37p0IlHhYfzX+12bE/OT/cc4XtvIjTYjsTEmiCyohKjEuGi+mpHKK7uKKa063Wn+jbtLiIuJ4PJpwV0jxhgzsFlQCWH3LpmEW+HpDw+eN19Dcwub88r4yswkW9/DGBNUFlRCWOrwWG6ck8yfPjvCidrGDvN98MVRauqbuXGuVX0ZY4LLgkqIu++KKdQ1tvDcnw91mGdjdgnDYiO5dMrIwBXMGGPaYUElxE1LiuPqGaP44yeHqG1oPmf/6cYWtu4p59pZyUTaSoTGmCCz30J9wH1XTOFkXRN/+uzIOfveLaigrrHFqr6MMSHBgkofcMH4YVw0cThPf3iQhuazlxx+fXcJiXHRXDRxRJBKZ4wxX/JrUBGRa0WkQEQKReThdvaLiDzu7M8WkQVe+xJEZJ2I7BWRfBG52EkfLiJbRWSf83OY1zGPOOcqEJFl/ry3QLvvismUVdezfteXi3idamjmnb0VXD87mfAwCWLpjDHGw29BRUTCgSeA5cBM4C4Rmdkm23IgzXndC6z22vcYsFlVpwNzgXwn/WFgm6qmAduc9zjnXgmkA9cCv3PK0C9cPjWR9DHxPPn+flqcRbze3lNOQ7Pbqr6MMSHDn08qFwKFqnpAVRuBNcCKNnlWAM+rx3YgQUSSRSQeWAI8A6Cqjap60uuY55zt54CbvdLXqGqDqh4ECp0y9AsinkW8DhytZUteGeCp+kpJGMT81GGdHG2MMYHhz6CSAhR5vS920rqSZxJQCTwrIrtE5GkRGezkGa2qpQDOz1GdnOssInKviGSKSGZlZWXP7ixIls9KZsKIWH73XiEn6xr5YF8l189JJsyqvowxIcKfQaW933RtlzPsKE8EsABYrarzgVqcaq5eXg9VfUpVM1Q1IzGxb01pEh4mfOfyyeS6qvnR+jyaWpQb59g098aY0OHPoFIMpHq9HwuUdDFPMVCsqp866evwBBmAchFJBnB+VnRyrn7l1gUpjI6P5vXdJUwYEcuslPhgF8kYY87wZ1DZAaSJyEQRicLTiL6hTZ4NwN1OL7BFQJWqlqpqGVAkItOcfEuBPV7H3ONs3wOs90pfKSLRIjIRT+P/Z365syCKjgjn25dNAuCGOWMQsaovY0zo8NvCG6raLCIPAFuAcOAPqponIquc/U8Cm4Dr8DSq1wHf9DrFg8ALTkA64LXvZ8BaEfkr4Ahwh3O+PBFZiyf4NAP3q+rZgzr6ia8tGkdpVT13Xzw+2EUxxpiziOo5zQ4DRkZGhmZmZga7GMYY06eIyE5VzWhvn42oN8YY4zMWVIwxxviMBRVjjDE+Y0HFGGOMz1hQMcYY4zMWVIwxxviMBRVjjDE+Y0HFGGOMzwzowY8iUgkc7sUpRgJHfVQcf7Dy9Y6Vr3esfL0TyuUbr6rtzsg7oINKb4lIZkejSkOBla93rHy9Y+XrnVAvX0es+ssYY4zPWFAxxhjjMxZUeuepYBegE1a+3rHy9Y6Vr3dCvXztsjYVY4wxPmNPKsYYY3zGgooxxhifsaDiEJFrRaRARApF5OF29ouIPO7szxaRBZ0dKyLDRWSriOxzfg4LdPlEJFVE3hWRfBHJE5HveR3zLyLiEpEs53VdoMvn7DskIjlOGTK90kPh85vm9flkiUi1iDzk7Avk5zddRP4sIg0i8nddOTbAn1+75Quh79/5Pr9Q+P519PkF5PvnU6o64F94ljveD0wCooDdwMw2ea4D3gQEWAR82tmxwC+Ah53th4GfB6F8ycACZzsO+MKrfP8C/F0wPz9n3yFgZDvnDfrn1855yvAM/Ar05zcKWAg86n3NEPr+dVS+UPn+tVu+EPr+dVg+f3//fP2yJxWPC4FCVT2gqo3AGmBFmzwrgOfVYzuQICLJnRy7AnjO2X4OuDnQ5VPVUlX9HEBVa4B8IKWH5fB5+To5b9A/vzZ5lgL7VbU3szD0qHyqWqGqO4CmbhwbsM+vo/KFyvfvPJ/f+QT982vDX98/n7Kg4pECFHm9L+bcL35Hec537GhVLQXPfy48f40EunxniMgEYD7wqVfyA051zx968Xjf2/Ip8JaI7BSRe73yhNTnB6wE/tQmLVCfX0+ODeTn16kgf//OJxS+f13hr++fT1lQ8ZB20tr2te4oT1eO7a3elM+zU2QI8DLwkKpWO8mrgcnAPKAU+FWQynepqi4AlgP3i8iSHpajI774/KKAm4CXvPYH8vPzx7Fd1etrhMD373xC4ft3/hP49/vnUxZUPIqBVK/3Y4GSLuY537HlrVUozs+KIJQPEYnE8x/6BVV9pTWDqparaouquoHf43lMD3j5VLX1ZwXwqlc5QuLzcywHPlfV8taEAH9+PTk2kJ9fh0Lk+9ehEPn+dcaf3z+fsqDisQNIE5GJzl8EK4ENbfJsAO4Wj0VAlfNIfL5jNwD3ONv3AOsDXT4REeAZIF9Vf+19QJs2g1uA3CCUb7CIxDnlGQx8xascQf/8vPbfRZuqhwB/fj05NpCfX7tC6PvXUflC5fvXGX9+/3wr2D0FQuWFp/fPF3h6afzASVsFrHK2BXjC2Z8DZJzvWCd9BLAN2Of8HB7o8gGX4XnUzgaynNd1zr7/dvJm4/mSJwehfJPw9IbZDeSF2ufn7IsFjgFD25wzkJ9fEp6/eKuBk852fAh9/9otXwh9/zoqX6h8/8737+v3758vXzZNizHGGJ+x6i9jjDE+Y0HFGGOMz1hQMcYY4zMWVIwxxviMBRVjjDE+Y0HFGD8RkVNBvPZ7IpIRrOubgcuCijHGGJ+xoGJMAInIjSLyqYjsEpG3RWS0k36517oYu0QkTkSSReQDJy1XRBY7ee8Sz/ofuSLy8+DekTFns6BiTGB9BCxS1fl4pkD/Byf974D7VXUesBg4DfwFsMVJmwtkicgY4OfAVXgmElwoIjcHsPzGnFdEsAtgzAAzFvhfZ96mKOCgk/4x8GsReQF4RVWLRWQH8AdnQsbXVDVLRK4C3lPVSgAn/xLgtUDfiDHtsScVYwLrt8B/qups4DtADICq/gz4NjAI2C4i01X1AzwBwwX8t4jcTfvTqBsTMuxJxZjAGoonSMCXM+AiIpNVNQfIEZGLgekichpwqervnRl0F+Cp+npMREYCJ/DMXvvbgN6BMedhQcUY/4kVkWKv97/Gs674SyLiArYDE519D4nIlUALsAd4E88U6X8vIk3AKeBu9SwX8AjwLp6nlk2q2tMp2Y3xOZul2BhjjM9Ym4oxxhifsaBijDHGZyyoGGOM8RkLKsYYY3zGgooxxhifsaBijDHGZyyoGGOM8Zn/Dz+EaqnoggKLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X_cancer, y_cancer, train_size=0.8, shuffle=False)\n",
    "regr = NormalLR()\n",
    "regr.fit(X_train, y_train)\n",
    "print(\"NormalLR\", mse(y_test, regr.predict(X_test)))\n",
    "regr = GradientLR(0.1, iterations=10000)\n",
    "regr.fit(X_train, y_train)\n",
    "print(\"GradientLR\", mse(y_test, regr.predict(X_test)))\n",
    "build_plot(X_train, y_train, X_test, y_test, iterations=10000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NormalLR 0.18996404389207405\n",
      "GradientLR 0.1894597185320024\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X_spam, y_spam, train_size=0.8, shuffle=False)\n",
    "regr = NormalLR()\n",
    "regr.fit(X_train, y_train)\n",
    "print(\"NormalLR\", mse(y_test, regr.predict(X_test)))\n",
    "regr = GradientLR(0.1, iterations=10000)\n",
    "regr.fit(X_train, y_train)\n",
    "print(\"GradientLR\", mse(y_test, regr.predict(X_test)))\n",
    "build_plot(X_train, y_train, X_test, y_test, iterations=10000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}