heartbeat_data_generator.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from mpl_toolkits.axes_grid1 import ImageGrid\n",
    "from pyts.image import GramianAngularField, MarkovTransitionField\n",
    "from pyts.datasets import load_gunpoint\n",
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "from sklearn.metrics.pairwise import pairwise_distances\n",
    "from sklearn.utils import resample\n",
    "from imblearn.combine import SMOTETomek\n",
    "from imblearn.over_sampling import SMOTE\n",
    "from imblearn.under_sampling import TomekLinks, NearMiss\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import torchvision\n",
    "from torchvision import datasets, transforms\n",
    "from torch.utils.data import DataLoader, Dataset,TensorDataset\n",
    "\n",
    "from matplotlib import colors\n",
    "from IPython import display\n",
    "import os\n",
    "from torch.utils.data.dataset import Dataset\n",
    "from matplotlib.pyplot import imread\n",
    "import glob\n",
    "import os\n",
    "import random\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "\n",
    "from torchvision.models import alexnet, vgg16, resnet152, resnet18, vgg19\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "from torch.distributions.utils import lazy_property\n",
    "import math\n",
    "import h5py\n",
    "from PIL import Image\n",
    "\n",
    "\n",
    "import torch.nn.functional as nnf\n",
    "from torch.utils.data import random_split\n",
    "from torch.optim import SGD \n",
    "from torch.distributions import constraints\n",
    "import torchvision as torchv\n",
    "import torchvision.transforms as torchvt\n",
    "import pickle\n",
    "from sklearn.preprocessing import StandardScaler, MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_hb_train = pd.read_csv('mitbih_train.csv', header=None)\n",
    "df_hb_test = pd.read_csv('mitbih_test.csv', header=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21892, 188)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hb_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    72471\n",
       "4.0     6431\n",
       "2.0     5788\n",
       "1.0     2223\n",
       "3.0      641\n",
       "Name: 187, dtype: int64"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hb_train[187].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    18118\n",
       "4.0     1608\n",
       "2.0     1448\n",
       "1.0      556\n",
       "3.0      162\n",
       "Name: 187, dtype: int64"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hb_test[187].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>178</th>\n",
       "      <th>179</th>\n",
       "      <th>180</th>\n",
       "      <th>181</th>\n",
       "      <th>182</th>\n",
       "      <th>183</th>\n",
       "      <th>184</th>\n",
       "      <th>185</th>\n",
       "      <th>186</th>\n",
       "      <th>187</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>80482</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.899749</td>\n",
       "      <td>0.641604</td>\n",
       "      <td>0.314536</td>\n",
       "      <td>0.101504</td>\n",
       "      <td>0.107769</td>\n",
       "      <td>0.111529</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>0.063910</td>\n",
       "      <td>0.058897</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80483</th>\n",
       "      <td>0.732252</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.983773</td>\n",
       "      <td>0.484787</td>\n",
       "      <td>0.178499</td>\n",
       "      <td>0.111562</td>\n",
       "      <td>0.081136</td>\n",
       "      <td>0.050710</td>\n",
       "      <td>0.046653</td>\n",
       "      <td>0.046653</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80484</th>\n",
       "      <td>0.991984</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.601202</td>\n",
       "      <td>0.280561</td>\n",
       "      <td>0.158317</td>\n",
       "      <td>0.146293</td>\n",
       "      <td>0.104208</td>\n",
       "      <td>0.090180</td>\n",
       "      <td>0.080160</td>\n",
       "      <td>0.070140</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80485</th>\n",
       "      <td>0.778378</td>\n",
       "      <td>0.528378</td>\n",
       "      <td>0.258108</td>\n",
       "      <td>0.128378</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.037838</td>\n",
       "      <td>0.167568</td>\n",
       "      <td>0.235135</td>\n",
       "      <td>0.227027</td>\n",
       "      <td>0.222973</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80486</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.914931</td>\n",
       "      <td>0.531250</td>\n",
       "      <td>0.303819</td>\n",
       "      <td>0.178819</td>\n",
       "      <td>0.157986</td>\n",
       "      <td>0.128472</td>\n",
       "      <td>0.126736</td>\n",
       "      <td>0.118056</td>\n",
       "      <td>0.111111</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81118</th>\n",
       "      <td>0.771529</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.776801</td>\n",
       "      <td>0.316344</td>\n",
       "      <td>0.165202</td>\n",
       "      <td>0.124780</td>\n",
       "      <td>0.117750</td>\n",
       "      <td>0.108963</td>\n",
       "      <td>0.084359</td>\n",
       "      <td>0.091388</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81119</th>\n",
       "      <td>0.590000</td>\n",
       "      <td>0.373333</td>\n",
       "      <td>0.283333</td>\n",
       "      <td>0.303333</td>\n",
       "      <td>0.340000</td>\n",
       "      <td>0.263333</td>\n",
       "      <td>0.136667</td>\n",
       "      <td>0.120000</td>\n",
       "      <td>0.090000</td>\n",
       "      <td>0.110000</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81120</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.939286</td>\n",
       "      <td>0.553571</td>\n",
       "      <td>0.285714</td>\n",
       "      <td>0.166071</td>\n",
       "      <td>0.141071</td>\n",
       "      <td>0.112500</td>\n",
       "      <td>0.108929</td>\n",
       "      <td>0.094643</td>\n",
       "      <td>0.096429</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81121</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.964981</td>\n",
       "      <td>0.778210</td>\n",
       "      <td>0.513619</td>\n",
       "      <td>0.214008</td>\n",
       "      <td>0.090791</td>\n",
       "      <td>0.112840</td>\n",
       "      <td>0.097276</td>\n",
       "      <td>0.075227</td>\n",
       "      <td>0.070039</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>81122</th>\n",
       "      <td>0.724014</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.801075</td>\n",
       "      <td>0.252688</td>\n",
       "      <td>0.093190</td>\n",
       "      <td>0.066308</td>\n",
       "      <td>0.043011</td>\n",
       "      <td>0.030466</td>\n",
       "      <td>0.021505</td>\n",
       "      <td>0.021505</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>641 rows × 188 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            0         1         2         3         4         5         6    \\\n",
       "80482  1.000000  0.899749  0.641604  0.314536  0.101504  0.107769  0.111529   \n",
       "80483  0.732252  1.000000  0.983773  0.484787  0.178499  0.111562  0.081136   \n",
       "80484  0.991984  1.000000  0.601202  0.280561  0.158317  0.146293  0.104208   \n",
       "80485  0.778378  0.528378  0.258108  0.128378  0.000000  0.037838  0.167568   \n",
       "80486  1.000000  0.914931  0.531250  0.303819  0.178819  0.157986  0.128472   \n",
       "...         ...       ...       ...       ...       ...       ...       ...   \n",
       "81118  0.771529  1.000000  0.776801  0.316344  0.165202  0.124780  0.117750   \n",
       "81119  0.590000  0.373333  0.283333  0.303333  0.340000  0.263333  0.136667   \n",
       "81120  1.000000  0.939286  0.553571  0.285714  0.166071  0.141071  0.112500   \n",
       "81121  1.000000  0.964981  0.778210  0.513619  0.214008  0.090791  0.112840   \n",
       "81122  0.724014  1.000000  0.801075  0.252688  0.093190  0.066308  0.043011   \n",
       "\n",
       "            7         8         9    ...  178  179  180  181  182  183  184  \\\n",
       "80482  0.071429  0.063910  0.058897  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "80483  0.050710  0.046653  0.046653  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "80484  0.090180  0.080160  0.070140  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "80485  0.235135  0.227027  0.222973  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "80486  0.126736  0.118056  0.111111  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "...         ...       ...       ...  ...  ...  ...  ...  ...  ...  ...  ...   \n",
       "81118  0.108963  0.084359  0.091388  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "81119  0.120000  0.090000  0.110000  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "81120  0.108929  0.094643  0.096429  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "81121  0.097276  0.075227  0.070039  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "81122  0.030466  0.021505  0.021505  ...  0.0  0.0  0.0  0.0  0.0  0.0  0.0   \n",
       "\n",
       "       185  186  187  \n",
       "80482  0.0  0.0  3.0  \n",
       "80483  0.0  0.0  3.0  \n",
       "80484  0.0  0.0  3.0  \n",
       "80485  0.0  0.0  3.0  \n",
       "80486  0.0  0.0  3.0  \n",
       "...    ...  ...  ...  \n",
       "81118  0.0  0.0  3.0  \n",
       "81119  0.0  0.0  3.0  \n",
       "81120  0.0  0.0  3.0  \n",
       "81121  0.0  0.0  3.0  \n",
       "81122  0.0  0.0  3.0  \n",
       "\n",
       "[641 rows x 188 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hb_train[df_hb_train[187] == 3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(87554, 188)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hb_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21892, 188)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_hb_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# omit last column as y\n",
    "x_train = df_hb_train.iloc[:,:187]\n",
    "x_test = df_hb_test.iloc[:,:187]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For exploratory purpose only\n",
    "gasf = GramianAngularField(image_size=187, method='summation')\n",
    "x_gasf_train = gasf.fit_transform(x_train)\n",
    "x_gasf_test = gasf.transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f4590e780b8>]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAD4CAYAAAD8Zh1EAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3deXScd33v8fd3Vm0zWkdeZcmLYsd2SOIozkYWmgUnhbgsCQlNgbIEelkLbUkPELhwbstSoJfbQEgpEFJIQoCCCw4mJQkESBzbsbN4izfZli1bm7UvM6P53T+eGXkkjTxjeWaeWb6vc3Q8eubRM78zlj766fv8FjHGoJRSKv857G6AUkqp9NBAV0qpAqGBrpRSBUIDXSmlCoQGulJKFQiXXS9cV1dnmpqa7Hp5pZTKS9u2besyxgQSPWdboDc1NbF161a7Xl4ppfKSiBye6TktuSilVIHQQFdKqQKhga6UUgVCA10ppQqEBrpSShWIpIEuIt8VkQ4ReWWG50VEviEi+0XkJRFZk/5mKqWUSiaVHvr3gXVneP5moDn6cTfwrXNvllJKqbOVNNCNMb8Hes5wynrgB8byHFAlIvPS1UClcs0vdhyjbyRkdzOUmiYdNfQFwNG4z9uix6YRkbtFZKuIbO3s7EzDSyuVXW2nhvnoIzv42QttdjdFqWnSEeiS4FjCXTOMMQ8YY1qMMS2BQMKZq0rltPa+0Un/KpVL0hHobUBD3OcLgeNpuK5SOUcDXeWydAT6BuAd0dEulwN9xpj2NFxXqZxzMhrkJzXQVQ5KujiXiDwMXAfUiUgb8FnADWCMuR/YCNwC7AeGgb/OVGOVstuJ/tFJ/yqVS5IGujHmziTPG+CDaWuRUjksPtCNMYgkuoWklD10pqhSZ+FEtNQSDEc4NaxDF1Vu0UBX6iyc6BulzOOceKxULtFAVypFkYihY2CUCxZUAnCif8TmFik1mQa6UinqGQ4SGjdctKgKgBN9Yza3SKnJNNCVSlGsxPKaBVWI6EgXlXs00JVKUSzQF1SXUlfh5USfllxUbtFAVypFsR75XH8Jc/0lnOjXkovKLRroSqXoZP8oTocQ8HmZW1mis0VVztFAVypF7X2jBCq8OB3CXH8J7VpyUTlGA12pFHUMjDHH7wVgbmUJ/aNhhsbCNrdKqdM00JVK0WhwnDKPtVpGQ00ZAEdPDdvZJKUm0UBXKkVj4XFK3NaPTGM00A93a6Cr3JF3gW6MoXc4aHczVBEaDUXwuqxp/0215QAc0UBXOSTvAv2bTx/g4i88wWho3O6mqCIzFh7HG+2hV5a5qSx1c7hnyOZWKXVa3gX6wupSjIEjPdozUtk1Fo7gdZ3+kWmsLdOSi8opeRfoS+oqADjYqT0jlV1WoDsnPl9Uo4GuckveBXpTnXUz6lCXBrrKrrHQ+LQe+rHeEULjERtbpdRpeRfovhI3AZ+XQ12DdjdFFZmxcGSihg7QWFvOeMRwvFcnGKnckHeBDrC4rlx76CqrwuMRwhEzqeSiQxdVrsnLQF+iga6yLBgtq0wuuVhDFw/rDXqVI/Iy0BfXldM1GKRvRPd0VNkxFpoe6PU+L16XgyPd2rlQuSFvAx2gVXvpKkvGwtFAd58uuTgcoiNdVE7Jy0BfErCGLmrZRWVLbCJbiXvyj8zC6lKO6U1RlSPyMtAX1ZThEDioga6yZKKHHndTFKzdi3SUi8oVeRnoHpeDhpoyDnTq0EWVHWNhq4ceX0MHmF9VyqnhEMNBXUZX2S8vAx2gud7H3hMDdjdDFYkZe+hVpQAcO6W9dGW/vA30lfN8HOwc1EW6VFZMjHJJUEMHaNOyi8oBeRvo58/zEzHw6kntpavMm6nksqDKmlykdXSVC/I60AF2t/fb3BJVDGYquQR8XlwO0ZKLygl5G+iLasoo9zjZ3a49dJV5M/XQnQ5hXlWJDl1UOSFvA93hEJbP9WkPXWXFTDV0sG6Mag9d5YKUAl1E1onIXhHZLyL3JHh+kYg8JSLbReQlEbkl/U2d7vx5fna392OMycbLqSI2U8kFrKGLWkNXuSBpoIuIE7gPuBlYCdwpIiunnPZp4MfGmIuBO4BvpruhiZw/z0//aJjjfaPZeDlVxGaaKQqwsKqUE/2jui66sl0qPfS1wH5jzEFjTBB4BFg/5RwD+KOPK4Hj6WvizCZujB7XsovKrFgP3eNMUHKpLiVi4IR2LJTNUgn0BcDRuM/bosfifQ64S0TagI3AhxNdSETuFpGtIrK1s7NzFs2drHmOtabLfp0xqjJsLDyOyyG4EgV6dOii3hhVdksl0CXBsalF6zuB7xtjFgK3AA+JyLRrG2MeMMa0GGNaAoHA2bd2Cn+Jm3qfl/0dGugqs8ZCkWkjXGLmV5UAOhZd2S+VQG8DGuI+X8j0ksp7gB8DGGOeBUqAunQ0MJmlgQpd00VlnLX93PQbomDdFAWd/q/sl0qgbwGaRWSxiHiwbnpumHLOEeB6ABE5HyvQz72mkoJl9RXs7xjUkS4qo8bC4zP20EvcTuoqvFpyUbZLGujGmDDwIWATsBtrNMtOEfm8iNwaPe0TwPtE5EXgYeBdJksJuzRQzsBomM6BsWy8nCpSY+GZSy5g3RjVQFd2c6VykjFmI9bNzvhj98Y93gVcld6mpWZZvQ+wbozW+0vsaELWRSKG1u4hFlaX4TlDyKj0sWroiUsuAAuqStijq38qm6UU6LlsWb010uVAxyBXLs1K2d4WkYjhly+38+Tukzyzr4vuoSCrF/j5zjsuZW5lcfwis9NYeDzhLNGYBVWlPLmnA2MMIonGESiVeXnfvZvj91LhdRX8SJd//e0+PvLwdp7Z18U15wX45LoVHOocYv19f2B/h/YMMy1pyaWqlNFQhO6hYBZbpdRked9DFxGWBso50Fm429Ftbe3h357cx5suXsBXb7sQh8PqAV63PMA7vvs8b/v2c/zwfZexYq4/yZXUbI2Gxin3zvzjsqA6Ohb91Ah1Fd5sNUupSfK+hw6wNDrSpRAd7h7iww9vZ2F1GZ9fv2oizMGaKfvI3ZfjcgofeGibTj3PoGQ9dB2LrnJBQQT64tpyTvSPMhIsrN2LWruGeNu3n2MkNM79d12Cr8Q97ZylgQr+6U0X0No9zI+3Hk1wFZUOVqDPfFN0oc4WVTmgIAJ9Ua31w3T01LDNLUkfYwz/8NOXGA2P8/D7Lmfl/JnLKX+2op6Wxmq+8dt9BfdLLVecaRw6gL/URYXXRZtOLlI2KoxAr7EC/Uh34QT6/+zu4PlDPXzipuUTi5DNRET4h3UrONk/xoPPtmalfcVmLBQ54ygXEWG+bnShbFYQgd5YWw7A4Z7CCPTweIQvPr6bJYFy7ri0IfkXAGsX13Dd8gDfevoAfSOhDLew+CQruYA10kVr6MpOBRHo1WVufF4XRwsk0B967jAHOof45LoVuBOs7jeTv7tpOX0jIR74/YEMtq44JSu5gM4WVfYriEAXERpqyjjcnf9DFzsHxvjab17l6uY6blo556y+dvWCSt544Xy++4dWBsfCGWph8THGJB3lAhCoKKF3OKSjjZRtCiLQARpryzhSAD30f9m0l9HwOJ+7ddWsZhy+fe0iRkLj/GFfVtZGKwqhcYMxzLjaYkxNhQeAU8M6uUjZo2ACfVFNGUdPjRCJ5O+qi92DY/xsextvX7uIpYGKWV2jpakaf4mL3+7uSHPritdY2Bo5lKyHXlNmBXqPzhZVNimcQK8tIxiOcHIgf7cB++kLbYTGDXdd3jjra7idDq5dXs9Tezvy+pdbLhkNRTeITtZDL48G+qAGurJH4QR6ng9dNMbw8PNHaWmspnmO75yudf2KeroGg7zY1pum1hW3VHvotdGSS4+WXJRNCibQG2vye+jiswe7OdQ1xJ1rF53zta5bHsAhaNklTWIbRCcL9GotuSibFUygz6sqwemQvBy6GIkY/mXTXuoqPNxywbxzvl5VmYfLFtey4cXjWnZJg7FYySXJOPTqMmtphm4tuSibFEygu50O5vrzc6bez7Yf44UjvXxy3QpKPWcOjVTdsbaBIz3D/OlAd1quV8wmSi5nmCkK4HI6qCpz6ygXZZuCCXSAgM+bd1vRDY6F+eLju1mzqIq3rFmYtuu+ftVcqsrcPLzlSNquWaxSLbmANdJF10RXdtFAt9n3/nCIrsEgn33j5KVxz1WJ28mbL17Ib3aeoHswv96TXHM60JP/9VRT7tFRLso2BRXo9XkW6H0jIf79mYPccP4cLmyoSvv171jbQGjc8KuX29N+7WIyFkptlAtYga4lF2WXggr0gM9L91Awb6Ze/8czB+kfDfO3NzZn5PrnzfGxpK6cJ3adzMj1i0Wsh16SpIYOVqBryUXZpaACvd5n7RqTD6MMXjnWx7d+d4BbL5zPqvmVGXudG1fN4dkD3boC4znoipas/Ak2GJmqptzDqaEgxujoIpV9BRXoAZ+1l2NHjs8WHQ6G+cjD26kt9/L59asy+lo3rZxDOGJ4eq+OSZ+tF470Mq+yhHp/SdJza8o9hCOG/lFdHE1lX0EFen000HO9jn7fU/s52DXE1952IVXRySiZclFDNXUVXi27nINtrT2saaxO6dyJ6f9adlE2KKhAP91Dz91Ab+8b4TvPHGL9RfO5cmldxl/P6RBuXFnPU3s6GA5qr/FsHe8d4XjfKC1nHei5+z2oCldBBXpdRe730L/2m1cxxtqMIlvevGYhQ8FxfvmSjnY5W1sPnwKgpbEmpfNPB7res1DZV1CB7nE5qC5z52wN/YldJ3lsWxvvuqqJhuhiYtnQ0ljNsvoKHn5eJxmdrW2tPZR5nJw/L7UF07SHruxUUIEO1kiXXOyhH+wc5OOP7uCCBZV8/MbzsvraIsIdlzaw/Ugve070Z/W188nu9n7+9tEdXPXFJ/neHw8xFh5n86EeLmqowpXiVoC15dZfiTp0Udmh4AI94PPmXA09PB7hI49sx+UU7v+rSyhJsq52JrxlzUI8Tgc/3tKW9dfOByPBcd727Wd5YtdJ6nxe/vd/72L1Zzex58QA154XSPk6pR4nJW4HpzTQlQ1cdjcg3ep9Xp5vza29Rb/zh0O8cqyfb/7lGhZUldrShupyD9cuD/D4K+18+s/PT+syA4Vg084T9I+GeeTuy7lscQ2PbWtjd3s/Vy6t4/oV9Wd1rdpyr/bQlS1S6qGLyDoR2Ssi+0XknhnOuV1EdonIThH5UXqbmbpYDz1XJnbs7xjg60+8yutXzeHm1XNtbcvNq+fS3jeqG18k8Ni2oyyqKWNtUw0iwu0tDXz2jau4ceWcs/7lV1Pu0WGLyhZJA11EnMB9wM3ASuBOEVk55Zxm4B+Bq4wxq4CPZaCtKQn4vATDkZyY2NE/GuLuh7bhK3HxhfWrZ7Xpczpdf/4c3E7h8VdO2NqOXNN2ylpm+C1rFqblL5fq6GxRpbItlR76WmC/MeagMSYIPAKsn3LO+4D7jDGnAIwxtk1LDExMLrJ3pMt4xPDxR3dwpHuY+96+JqVZhplWWermyqV1PP5Ke878BWO3oz3DfOq/XsEYePOaBWm5Zq2u56JskkqgLwCOxn3eFj0W7zzgPBH5o4g8JyLrEl1IRO4Wka0isrWzs3N2LU4itp6LnTdGjTF8bsNO/md3B/e+cSWXLam1rS1T3bx6Lkd7Rnj5WJ/dTbFd26lhbvr673n+UA+f/vPz0zaUVEsuyi6pBHqiv0Gndu9cQDNwHXAn8B0RmbYerDHmAWNMizGmJRBIfeTA2QjkwPT/bz59gIeeO8z7r1nCO65osq0didy8eh4el4OfbNPRLr97tZOR0Dg//Zsree/VS9J23ZpyD8PBcUajy+4qlS2pBHob0BD3+ULgeIJzfmGMCRljDgF7sQI+6+r99gb6Y1uP8pVNe/mLi+bzyXUrbGnDmVSWubl59Vx+vv1Y0QfO5oM91Pu8KU8aSpWu56LskkqgbwGaRWSxiHiAO4ANU875OfA6ABGpwyrBHExnQ1Pl87rwuhy2lFwef7mde372Mq9dVseX33phzg4NvL2lgf7RMJt2Fu/NUWMMmw91c9mS2rTfrNZAV3ZJGujGmDDwIWATsBv4sTFmp4h8XkRujZ62CegWkV3AU8DfG2Ns2Z1YRGzZim7jy+186OHtXNRQxbfuWoMnhd1t7HLFkloaakp5dMvR5CcXqMPdw5zsH+Oyxamt0XI2NNCVXVKaWGSM2QhsnHLs3rjHBvh49MN29T5vVtdzid0EXT3fz4PvXkuFN7fnazkcwm2XNPC1J17lSPcwi2qzt65Mrth8yOpvXL5EA10VjtztRp6DbPfQ95wYoGNgjLsub8z5MI956yULEbEm1BSjzQd7qKvwsDRQkfZr12qgK5sUZKDX+0qyWkN/Zp81BPPq5syM3MmE+VWlXNMc4Cfb2hiPFNeY9C2tPfzq5Xaubg5kZLKXv8SN0yEa6CrrCjLQAz4vvcMhxsLZGcXxzL4uzptTwdxK+ycPnY3bWxpo7xvlqT3Fsz3dgc5B3vvgVhZUlfKZN6xM/gWz4HAI1WVunVyksq4gAz22FV1XFjaLHglaS6xek0e985gbVtbTUFPKJx57kZfbimOi0f/51W6MMTz47rUTte5MqNHp/8oGBRno2Zxc9HxrD8FwhKvPYonVXOF1OfnRey/HV+Li7d95jt3thb1W+gtHTvHkng7ef+3SjG8wUl2ms0VV9hVkoE9M/+/P/EiXFw6fwiGwtin9oyWyoaGmjEfuvpxyj4t3fe95jvWO2N2kWQmNR/jDvi6C4ci05472DLPhxeP888bd1JZ7eNeVTRlvT22Fh27dtUhlWX4MyThLEz30wcz/QB3qGmJBdSmlnuxvWpEuC6vL+P67L+W2+59l3b/+nttbGvjwny2jqixzJYmpRkPj7Ds5iAgsriunPMFooWA4wr8/c5DqMg93rm3gQOcQu9r7WTnPx72/2MmfDnTTVFvGGy+cT0f/GFc117GwupR3fvd5BqKrb37mDSsTXjvdaso9nBrWfUVVdhVkoNdWeBCBjv7MB3pr9xBNteUZf51MWzHXz2MfuIJvPnWAB//UyrbDp/jR+y6jzJOZb5FTQ0G+vGkvC6tLed3yej74oxc41GVtTOJ2CpcvqeWj1zfTMTDGFx/fwxy/l8Gx8Ymy0A+ebWVfx+DECB2P08FHr2/m16+c4P89uR9/iYtHtx5FBBqqy3joPZdRVeqmMUtj7mvKPJwaDjIeMThzdMawKjwFGehup4OaMk/Ge+jGGA51DvGmNC27arcVc/18486L+fPXzONv/nMbb/3Ws5R5nLxuRT0ffN2yWV3z16+088PNR/jAtUvpGQryg2db8ZW4eeVYH91DVuB9ZdNeaso9fO32CynzONl+pJef7zjGW+9/FoCV8/yMhSMMjIa4/65L6Boc4+tPvMpdly3i1osW8OLRXlqaqnnNwio+dkMzwfEILoeDn73QxlN7O/jMG1YyrzK7O0XVlHswBnqHg9RWeLP62qp4FWSgQ3Tnogz30LuHggyMhQuihx7v9avm8sU3v4b7f3+AiLEC1yHCiugiVtedF+DZg9188fE9/NObLmD1gsqE1/nT/i4+/PB2jLGGdgIsCZQzODZOQ00Z3/vrSxkcDfPzHcf5X9edvlG5bvU8PnpDM9/7Yysi8L6rl+CesknzXZc3Tjy+pLF64rGI4HVZ5a/bWhq4raUBO9REQ/yUBrrKooIO9Ez30FujJYLFgcIKdIDbL23g9ksbiEQMH354O1/69Z6J59ZfNJ8nd3cwMBbm/Q9t4yd/cwVHe0aYV1lCOGL4+8deZNuRUxgD582p4D/fexmbdp6kwuvk1gsXTCtBJFovvszjmvVfBbmguswNQM+Q1tFV9hRsoNf7SjjQ0ZXR1zgYC/QC66HHcziEr95+IZc0VrOsvoI/Huji2787yFx/CV+57UI+8sh2rvjnJyfOdzqEco+T91+zlHKPk7dd2kC9r4S/iutRF4PKUivQ+0Y00FX2FGygx3roxpiM7eXZ2jWEyyEsrM5ufTbbStxO3v3axQBcc16Aa5oDLKopo6GmjH+782KeO9jD2sU1HO8d4VjvCO+9enHWa9a5Jhbo/RroKosKNtDrfV5C44be4RDVGZoR2No9RENNGS5nQQ7nn9FVy+omHt+0ai43rZprY2tyk79Ee+gq+wo2ibIxFv1Q1zCL6wq33KJmz68lF2WDgg302HoumRrpYoyhtaswxqCr9HM6BJ/XRf+oBrrKnoIN9NM99MxM/2/vG2UkNF6QI1xUevhL3dpDV1lVsIFe74+t55KZHnpsxuL5c9O7wbAqHP5SN/0jYbuboYpIwQZ6ucdJqduZsRUXdx7vRwRWzPNn5Poq/1WWunSUi8qqgg10EaHe783YzkW7jvfTVFueN1vOqezzl2jJRWVXwQY6QKAic3uLWqv8ae9czayy1K03RVVWFXSgWz309N8U7R8NcaRnmJXzNdDVzCr1pqjKsoIO9Ez10Pe0DwBoD12dkb/UzXBwnND49E03lMqEgg70en8J/aNhRkPp3Sx613Fr/03toasz0en/KtsKOtADFZnZW3RXez91FZ6JyUtKJeIvtW6Ya9lFZUthB7o/Ols0jYFujOGP+7u5qKEqY4t+qcKgKy6qbCvsQM9AD/2FI70c6x3h5tXz0nZNVZgmSi6jOrlIZUdBB3q9Pxbo6Rvp8suXjuNxOrhx1Zy0XVMVJl1xUWVbQQd6bbkXj8tBa/dwWq4XiRg2vtzOtcsDEz+sSs1Eb4qqbCvoQHc6hIsbqth8qDst19vS2sPJ/jHe8Bott6jkdAldlW0FHegAVyytZefxfnqHg+d8rd/u6cDtFK4/X8stKrkStxOPy6E9dJU1KQW6iKwTkb0isl9E7jnDeW8VESMiLelr4rm5cmkdxsDmQz0pnT8wGuKfN+7mnp++xEPPthKOmxTy9N4O1i6u0fVbVMp0+r/KpqTJJCJO4D7gRqAN2CIiG4wxu6ac5wM+AmzORENn68KGSkrcDp49YA01BJgTXVp3qn0nB/jr72/heO8I1WUeHtlyFIdD+MvLGjneO8KrJwe57ZKGbDZf5Tl/iUtLLiprUumhrwX2G2MOGmOCwCPA+gTnfQH4MpCZHSVmyety0tJYw69ebuf6r/6O9z64dcZzf/DsYboHgzz2gSvY+ukbuKihivt/d4DweISn93YCcN3yQLaargqArueisimVQF8AHI37vC16bIKIXAw0GGN+eaYLicjdIrJVRLZ2dnaedWNn64qltXQOjBEaj/DysT66ZthntL1vlMbaMi5prEFE+ODrlnG0Z4Rf7DjOU3s7WFBVyrL6iqy1W+W/ygxscjEcDPOLHcfSek1VGFIJ9ETTIc3EkyIO4OvAJ5JdyBjzgDGmxRjTEghkr6d71+WNfOYNK/neuy4F4E8HEo96Odk/Oqkcc/2KepbP8fGJx17kiV0nuXZ5QGeHqrOSiW3ofrT5CB99ZAdHe9IzHFcVjlTu7rUB8YXjhcDxuM99wGrg6WjYzQU2iMitxpiZ6xtZVFnq5j2vXcx4xOArcfGn/V3ceuH8aeed6B/l/Hmnt5RzOIT/e+dF/HZ3ByLwposXTPsapc4kEzdFt7aeAmAsnN5F51T+SyXQtwDNIrIYOAbcAbw99qQxpg+oi30uIk8Df5crYR7P6RCuXFrLM/u6MMZM6m2HxiN0DY4xd8oN0xVz/ayYq6sqqtmxSi4hIhGDw3Huf90ZY9h62Ar00LhJcrYqNklLLsaYMPAhYBOwG/ixMWaniHxeRG7NdAPT7apldRzrHeHIlD9XuwbHMAbmVCYeAaPUbPhL3EQMDAbTU0c/0jM8cQ8orIGupkhpQLUxZiOwccqxe2c497pzb1bmXLnU+mNi86EeGmvLJ46f6LMG50ztoSt1LuKn/6djuYht0d45QFA3zlBTFPxM0akaa8sQgeO9I5OOn+y3An2mMepKzUa610TfGhfoYQ10NUXRBbrb6aC23DNtjfSJHrqWXFQapXM9F2MMW1t78EVnKmsNXU1VdIEOEPCV0NE/ef7Tif4x3E6hpsxjU6tUITpdcjm3GvqpoSAf+M9tvHpykGujk9tCEe2hq8mKMtDrfd5pPfST/aPU+0rSMhJBqZhY3fxcF+j63H/v5Mk9HfzjzSu4+5olgN4UVdMVZaDP8XsnauYxJ/pGtdyi0q6y7NxLLuMRw1N7OviLixbw/muX4nFZP7YhraGrKYoy0Ot9JXQNBhmPnO7hnOwf1REuKu0qPC4cwjlNLnqprZf+0TDXnGeVWtxODXSVWFEG+hy/l/GIoXvIKrsYYzgxZdq/UungcAi+knOb/v/7V7sQgdcus4bcuh2xQNeSi5qsKAM94LOCu6PfCvTBsTDDwXHmVnrtbJYqULHZorP1zL5OXrOgkupy64a922Xd59Fhi2qqogz02ObRHdHNo3UMusokf+ns10TvHw2x/WgvVzefXszOFeuhR7SHriYrykCPBXeshx77t96nga7S71zWRN/WeorxiOGqZRPLJeF2Wj30UFh76Gqyogz0QEWsh24FeWd0bYyAT0suKv2sFRdnNw5978kBAFbOP71AXOymaFjHoaspijLQPS4HNeWeiVJL54AGusoc/zncFN13cpA5fu/EBCUAV6yHrjdF1RRFGegweXJR58AYHpcDf4lu/qzS71xKLvs6Bmiu9006dnqUi/bQ1WRFG+gBn3di+n/nwBiBCq/uRqQywl/qJhiOMBo6uw0pIhHD/o5BmudM3vbQ4RCcDtGZomqaog30Of6SSTV0LbeoTPGXzm76/7HeEYaD49N66AAuh2gPXU1TtIFe7/PSOTBGJGKsHroGusqQiQW6znK26L4O64bo1B46gMfp0Bq6mqZoA31BdSnhiKG9f5TOgTHqNdBVhlTOcgndfScHAWiunx7oLqfoKBc1TdEG+rKA9UOyp72fnuGg9tBVxsRutp9toL96cpCAz0tVgiWdXU6HllzUNMUb6NFez+ZDPRijQxZV5sx2TfT9HQOcl6DcAlpyUYkVbaDXVnipKffwpwNdwOnJRkql22x3LTrcM0xT3L638VxOvSmqpivaQAer7LLzeD+gPXSVObOpoTdULZ8AAAzrSURBVEcihr6RELXliXfQcjsdOmxRTVPcgT6nAhP9mdBAV5nidjrwlbjoHhxLfnLUwGgYY0737qfSYYsqkeIO9MDp+mSdllxUBgUqvHQNBlM+v3fEOjfRDVGwfklooKupijrQY+N7/SUuStxOm1ujClldhXdiEbhUxMozVTP00N1OIazL56opijrQYyNdtNyiMq3O5zmrkkvvsBXosT1Jp3I5HQR1+Vw1RVEH+lx/CRVelwa6yri6sy65nLmH7nE6tIeupinq5QVFhFsvmk9DdZndTVEFrrbcS99IiGA4gseVvB8VK7nM3EMXwqPaQ1eTFXWgA/zTmy6wuwmqCNT5rJub3UNjzKssTXp+37DVm6+ccZSLg6AOW1RTFHXJRalsiY2i6hpIrezSOxyi1O3E60p8s97jEt0kWk2jga5UFkwEeoo3RvtGQlTNUG4Bq4euwxbVVCkFuoisE5G9IrJfRO5J8PzHRWSXiLwkIr8Vkcb0N1Wp/BVbWiLVoYu9I6EZyy0QG4euJRc1WdJAFxEncB9wM7ASuFNEVk45bTvQYox5DfAT4MvpbqhS+Wyihp7iSJe+4WSBrsvnqulS6aGvBfYbYw4aY4LAI8D6+BOMMU8ZY4ajnz4HLExvM5XKb2UeF2UeZ/pKLk7RHrqaJpVAXwAcjfu8LXpsJu8BHk/0hIjcLSJbRWRrZ2dn6q1UqgDUVnhSDvTekSBVpYmn/YNO/VeJpRLoiXZOTtg1EJG7gBbgK4meN8Y8YIxpMca0BAKB1FupVAGwJhelGOjDoRnHoIMGukoslUBvAxriPl8IHJ96kojcAHwKuNUYk/ocZ6WKRF2FN6Vhi6OhccbCkTPW0F0O0eVz1TSpBPoWoFlEFouIB7gD2BB/gohcDHwbK8w70t9MpfJfqj30iYW5kvTQwxGDMRrq6rSkgW6MCQMfAjYBu4EfG2N2isjnReTW6GlfASqAx0Rkh4hsmOFyShWtQIWHnuFg0glBEwtzJRnlAuiNUTVJSlP/jTEbgY1Tjt0b9/iGNLdLqYJT5/NiDPQMBan3l8x43umlc898UxQgHIng0fmBKkq/E5TKkvPm+ADY0nrqjOf1Dsc2tzjTsEXrRzcU1h66Ok0DXaksubSphtpyD4+/0n7G82JL56ZUctHJRSqOBrpSWeJ0CDetmsNTezoYDY3PeF5/kqVzIa7kojV0FUcDXaksWrd6HkPBcZ7Z1zXjOb3DIZwOweed+RaXyxG7Kao9dHWaBrpSWXTFklr8Ja4zll16R4L4S1yIJJrTZ4ltkqGBruJpoCuVRR6Xg2uX1/OHfV0zjiHvGwlTVTbzCBewls8FHbaoJtNAVyrL1i6uoWNgjCM9wwmf7x0OnvGGKFiLc4H20NVkGuhKZdnaphoAnj/Uk/D5ZCstgrVJNKAbRatJNNCVyrLm+goqS91saZ050LWHrmZDA12pLHM4hEubqmecYNQ7HKIqSaDHhi1qoKt4GuhK2eDSphoOdQ3RMTA66XgkYugfDVGZ5KaoruWiEtFAV8oGly626uhbp/TSB0bDGHPmWaJwepRLsoW+VHHRQFfKBqvm+3EI7DkxMOl470h0HZeUSy7aQ1enaaArZQOvy0lDTRkHOgcnHY8tnZtslItbb4qqBDTQlbLJ0kAFBzomB3pfCgtzwenVFsO6OJeKo4GulE2WBso51DVEJG4seW8KuxWB3hRViWmgK2WTJYEKxsIRjvWOTBzri66FXnmGzS1Ahy2qxDTQlbLJ0kAFwKQ6eqolF10+VyWiga6UTZYGygE40Dk0cax3OESZxzmxmuJMdKaoSkQDXSmb1JR7qCx1c3BKDz3ZkEUAt662qBLQQFfKJiLC0kD5pJJL70gIfyqBHu2h68QiFU8DXSkbLQ1UTCq59A0nX2kRrO3sQEsuajINdKVstCRQQefAGP2j1s1Qq+Ry5hEuYPXuPU4HIV0+V8XRQFfKRovrygA43GVtdtE7Ekyphw7WjdFQWHvo6jQNdKVs1FRnjXRp7bbKLr3DyddCj3E5RDe4UJNooCtlo8aaaKB3DTEaGmcsHKEyxR66x+XQGrqaRANdKRuVepzM9ZfQ2j08MakolRo6WEvoaqCreBroStmssbaM1u4hugbHgOTruMS4XaIzRdUkGuhK2WxxXTmHu4d44bC12cWq+f6Uvs7tcBDUHrqKo4GulM2a6srpGgzy650nWFhdyqKaspS+zuXUHrqaTANdKZs11VoB/sf93Vy1tA4RSenr3E6HroeuJkkp0EVknYjsFZH9InJPgue9IvJo9PnNItKU7oYqVahiQxcBrmquS/nrXE4HQe2hqzhJA11EnMB9wM3ASuBOEVk55bT3AKeMMcuArwNfSndDlSpUsaGLAFcurU3569wO0bVc1CSuFM5ZC+w3xhwEEJFHgPXArrhz1gOfiz7+CfBvIiLGGO0+KJVEbOhiVZmbugpvyl/ndjp44cgpbvza7zLYOpUJH7m+mTdeOD/t100l0BcAR+M+bwMum+kcY0xYRPqAWqAr/iQRuRu4G2DRokWzbLJShefvX7+c6vLUhivG/NUVjWf9NSo3pDob+GylEuiJ7tBM7Xmncg7GmAeABwBaWlq0965U1FsuWXjWX3PLBfO45YJ5GWiNylep3BRtAxriPl8IHJ/pHBFxAZVATzoaqJRSKjWpBPoWoFlEFouIB7gD2DDlnA3AO6OP3wo8qfVzpZTKrqQll2hN/EPAJsAJfNcYs1NEPg9sNcZsAP4DeEhE9mP1zO/IZKOVUkpNl0oNHWPMRmDjlGP3xj0eBW5Lb9OUUkqdDZ0pqpRSBUIDXSmlCoQGulJKFQgNdKWUKhBi1+hCEekEDs/yy+uYMgs1B2kb00PbmB7axvTIhTY2GmMCiZ6wLdDPhYhsNca02N2OM9E2poe2MT20jemR623UkotSShUIDXSllCoQ+RroD9jdgBRoG9ND25ge2sb0yOk25mUNXSml1HT52kNXSik1hQa6UkoViLwL9GQbVttBRBpE5CkR2S0iO0Xko9HjnxORYyKyI/pxi83tbBWRl6Nt2Ro9ViMiT4jIvui/1Ta1bXnc+7RDRPpF5GO58B6KyHdFpENEXok7lvB9E8s3ot+fL4nIGpva9xUR2RNtw3+JSFX0eJOIjMS9n/dnun1naOOM/7ci8o/R93CviLzexjY+Gte+VhHZET1uy/uYlDEmbz6wlu89ACwBPMCLwMocaNc8YE30sQ94FWtD7c8Bf2d3++La2QrUTTn2ZeCe6ON7gC/lQDudwAmgMRfeQ+AaYA3wSrL3DbgFeBxrF6/Lgc02te8mwBV9/KW49jXFn2fze5jw/zb6s/Mi4AUWR3/mnXa0ccrzXwXutfN9TPaRbz30iQ2rjTFBILZhta2MMe3GmBeijweA3Vj7rOaD9cCD0ccPAn9hY1tirgcOGGNmO5M4rYwxv2f6DlwzvW/rgR8Yy3NAlYhkdJ+4RO0zxvzGGBOOfvoc1k5jtpnhPZzJeuARY8yYMeYQsB/rZz+jztRGERHgduDhTLfjXORboCfasDqnglNEmoCLgc3RQx+K/tn7XbvKGXEM8BsR2RbdsBtgjjGmHaxfTEC9ba077Q4m/+Dk0nsYM9P7lovfo+/G+qshZrGIbBeR34nI1XY1KirR/20uvodXAyeNMfvijuXS+wjkX6CntBm1XUSkAvgp8DFjTD/wLWApcBHQjvUnm52uMsasAW4GPigi19jcnmnE2ubwVuCx6KFcew+TyanvURH5FBAGfhg91A4sMsZcDHwc+JGI+G1q3kz/tzn1HkbdyeRORi69jxPyLdBT2bDaFiLixgrzHxpjfgZgjDlpjBk3xkSAfycLfzaeiTHmePTfDuC/ou05GSsJRP/tsK+FgPXL5gVjzEnIvfcwzkzvW858j4rIO4E3AH9pooXfaBmjO/p4G1Z9+jw72neG/9uceQ9hYuP7NwOPxo7l0vsYL98CPZUNq7MuWl/7D2C3MeZrccfja6dvAl6Z+rXZIiLlIuKLPca6afYKkzf4fifwC3taOGFSTyiX3sMpZnrfNgDviI52uRzoi5VmsklE1gGfBG41xgzHHQ+IiDP6eAnQDBzMdvuirz/T/+0G4A4R8YrIYqw2Pp/t9sW5AdhjjGmLHcil93ESu+/Knu0H1iiCV7F+I37K7vZE2/RarD8JXwJ2RD9uAR4CXo4e3wDMs7GNS7BGDrwI7Iy9d0At8FtgX/TfGhvbWAZ0A5Vxx2x/D7F+wbQDIaze43tmet+wygX3Rb8/XwZabGrffqw6dOz78f7ouW+J/v+/CLwAvNHG93DG/1vgU9H3cC9ws11tjB7/PvCBKefa8j4m+9Cp/0opVSDyreSilFJqBhroSilVIDTQlVKqQGigK6VUgdBAV0qpAqGBrpRSBUIDXSmlCsT/B6MWCvbU8vcjAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot 1 heartbeat\n",
    "x_train.values[87553, :]\n",
    "plt.plot(np.arange(187), x_train.values[87553, :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f816db8ef98>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "# GAF for 1 heartbeat\n",
    "plt.imshow(x_gasf_train[87553], cmap='gray', origin='lower')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f264c5a8470>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.imshow(x_gasf_test[21891], cmap='gray', origin='lower')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def recurrence_plot(s, eps=None, steps=None):\n",
    "    result = []\n",
    "    if eps==None: eps=0.1\n",
    "    if steps==None: steps=10\n",
    "    d = pairwise_distances(s[:, None])\n",
    "    d = d / eps\n",
    "    d[d > steps] = steps\n",
    "    return d/5. - 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "result = recurrence_plot(x_train.values[87553], steps=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(187, 187)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f8176c30748>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1008 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Recurrence plot for 1 heartbeat\n",
    "fig = plt.figure(figsize=(15,14))\n",
    "ax = fig.add_subplot(2, 3, 1)\n",
    "ax.imshow(result, cmap='gray')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(187, 187)\n",
      "(187, 187)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[-1.        ],\n",
       "       [-0.66442955],\n",
       "       [ 0.14093959],\n",
       "       ...,\n",
       "       [-1.        ],\n",
       "       [-1.        ],\n",
       "       [-1.        ]])"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1008 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Test\n",
    "fig = plt.figure(figsize=(15,14))\n",
    "ax = fig.add_subplot(2, 3, 1)\n",
    "ax.imshow(recurrence_plot(x_train.values[25][:, None], steps=10), cmap='gray')\n",
    "d = recurrence_plot(x_train.values[25][:, None], steps=10)\n",
    "scaler = MinMaxScaler((-1, 1))\n",
    "x = scaler.fit_transform(d.reshape(187 * 187, 1))\n",
    "x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train = df_hb_train[187].values.astype(int)\n",
    "y_test = df_hb_test[187].values.astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# upsampling to reduce class imbalance\n",
    "smt2 = SMOTE(sampling_strategy={1: 30000, 2: 20000, 3: 20000, 4: 10000}, n_jobs=-1, random_state=42)\n",
    "x_train_2, y_train_2 = smt2.fit_sample(x_train, y_train)\n",
    "\n",
    "gasf = GramianAngularField(image_size=150, method='summation')\n",
    "x_gasf_train_2 = gasf.fit_transform(x_train_2)\n",
    "x_gasf_test_2 = gasf.transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create placeholders, then populate with images\n",
    "with h5py.File('hb_data_mit_bih_balanced_2.hdf5', mode='w') as hdf5_file:\n",
    "    hdf5_file.create_dataset(\"x_train\", (len(x_gasf_train_2), 224, 224), np.float32)  # 224, 224 for resnet\n",
    "    hdf5_file.create_dataset(\"y_train\", (len(y_train_2),), np.int32)\n",
    "    hdf5_file.create_dataset(\"x_test\", (len(x_gasf_test_2), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_test\", (len(y_test),), np.int32)\n",
    "    \n",
    "    transform = transforms.Compose([transforms.Resize((224, 224))])\n",
    "    \n",
    "    for i, (x, y) in enumerate(zip(x_gasf_train_2, y_train_2)):\n",
    "        image = Image.fromarray(x.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_train\"][i, ...] = t\n",
    "        hdf5_file[\"y_train\"][i] = y\n",
    "\n",
    "    for i, (x, y) in enumerate(zip(x_gasf_test_2, y_test)):\n",
    "        image = Image.fromarray(x.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_test\"][i, ...] = t\n",
    "        hdf5_file[\"y_test\"][i] = y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# upsampling to try with GramianAngularField method='difference'\n",
    "# results are almost same for both\n",
    "smt3 = SMOTE(sampling_strategy={1: 30000, 2: 20000, 3: 20000, 4: 10000}, n_jobs=-1, random_state=42)\n",
    "x_train_3, y_train_3 = smt3.fit_sample(x_train, y_train)\n",
    "\n",
    "gasf = GramianAngularField(image_size=150, method='difference')\n",
    "x_gasf_train_3 = gasf.fit_transform(x_train_3)\n",
    "x_gasf_test_3 = gasf.transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "with h5py.File('hb_data_mit_bih_balanced_3.hdf5', mode='w') as hdf5_file:\n",
    "    hdf5_file.create_dataset(\"x_train\", (len(x_gasf_train_3), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_train\", (len(y_train_3),), np.int32)\n",
    "    hdf5_file.create_dataset(\"x_test\", (len(x_gasf_test_3), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_test\", (len(y_test),), np.int32)\n",
    "    \n",
    "    transform = transforms.Compose([transforms.Resize((224, 224))])\n",
    "    \n",
    "    for i, (x, y) in enumerate(zip(x_gasf_train_3, y_train_3)):\n",
    "        image = Image.fromarray(x.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_train\"][i, ...] = t\n",
    "        hdf5_file[\"y_train\"][i] = y\n",
    "\n",
    "    for i, (x, y) in enumerate(zip(x_gasf_test_3, y_test)):\n",
    "        image = Image.fromarray(x.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_test\"][i, ...] = t\n",
    "        hdf5_file[\"y_test\"][i] = y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "with h5py.File('hb_data_mit_bih_balanced_rp.hdf5', mode='w') as hdf5_file:\n",
    "    hdf5_file.create_dataset(\"x_train\", (len(x_train_2), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_train\", (len(y_train_2),), np.int32)\n",
    "    hdf5_file.create_dataset(\"x_test\", (len(x_test), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_test\", (len(y_test),), np.int32)\n",
    "    \n",
    "    transform = transforms.Compose([transforms.Resize((224, 224))])\n",
    "    \n",
    "    for i, (x, y) in enumerate(zip(x_train_2.values, y_train_2)):\n",
    "        r = recurrence_plot(x, steps=10)\n",
    "        image = Image.fromarray(r.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_train\"][i, ...] = t\n",
    "        hdf5_file[\"y_train\"][i] = y\n",
    "\n",
    "    for i, (x, y) in enumerate(zip(x_test.values, y_test)):\n",
    "        r = recurrence_plot(x, steps=10)\n",
    "        image = Image.fromarray(r.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_test\"][i, ...] = t\n",
    "        hdf5_file[\"y_test\"][i] = y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generating mixed hdf5: containing RP, GAF and mean of RP and GAF in 3 channels\n",
    "# Previous ones were single channel images\n",
    "with h5py.File('/media/asif/data/hb_data_mit_bih_balanced_mixed.hdf5', mode='w') as hdf5_file:\n",
    "    hdf5_file.create_dataset(\"x_train\", (len(x_train_3), 3, 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_train\", (len(y_train_3),), np.int32)\n",
    "    hdf5_file.create_dataset(\"x_test\", (len(x_test), 3, 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_test\", (len(y_test),), np.int32)\n",
    "    \n",
    "    transform = transforms.Compose([transforms.Resize((224, 224))])\n",
    "    \n",
    "    for i, (x1, x2, y) in enumerate(zip(x_gasf_train_3, x_train_3.values, y_train_3)):\n",
    "        image = Image.fromarray(x1.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        r = recurrence_plot(x2, steps=10)\n",
    "        image = Image.fromarray(r.astype(float))\n",
    "        t2 = np.array(transform(image))\n",
    "        hdf5_file[\"x_train\"][i, 0, ...] = t\n",
    "        hdf5_file[\"x_train\"][i, 1, ...] = t2\n",
    "        hdf5_file[\"x_train\"][i, 2, ...] = (t + t2) / 2\n",
    "        hdf5_file[\"y_train\"][i] = y\n",
    "\n",
    "    for i, (x1, x2, y) in enumerate(zip(x_gasf_test_3, x_test.values, y_test)):\n",
    "        image = Image.fromarray(x1.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        r = recurrence_plot(x2, steps=10)\n",
    "        image = Image.fromarray(r.astype(float))\n",
    "        t2 = np.array(transform(image))\n",
    "        hdf5_file[\"x_test\"][i, 0, ...] = t\n",
    "        hdf5_file[\"x_test\"][i, 1, ...] = t2\n",
    "        hdf5_file[\"x_test\"][i, 2, ...] = (t + t2) / 2\n",
    "        hdf5_file[\"y_test\"][i] = y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Hearbeat values range from 0 to 1; we will divide into quantiles and see which bin each value belongs to\n",
    "# and what the probability is to transfer from 1 bin to other\n",
    "# i.e if 1 value is in bin 1 and the next value is in bin 5, what is the probability of transferring from\n",
    "# bin 1 to 5\n",
    "# correlation matrix of probabilities is markov transition field\n",
    "def get_quantiles(min_value=0, max_val=1, k=10):\n",
    "    c = (max_val - min_value)/k\n",
    "    b = min_value + c\n",
    "    d = []\n",
    "    for i in range(1, k):\n",
    "        d.append(b)\n",
    "        b += c\n",
    "    d.append(max_val)\n",
    "    return d\n",
    "\n",
    "quantiles = get_quantiles()\n",
    "\n",
    "def value_to_quantile(x):\n",
    "    for i, k in enumerate(quantiles):\n",
    "        if x <= k:\n",
    "            return i\n",
    "    return 0\n",
    "\n",
    "# qv_train = np.vectorize(value_to_quantile)(x_train_3.values)\n",
    "# qv_test = np.vectorize(value_to_quantile)(x_test.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_mtf(x, size=10):\n",
    "    q = np.vectorize(value_to_quantile)(x)\n",
    "    r = np.zeros((q.shape[0], q.shape[0]))\n",
    "    y = np.zeros((size, size))\n",
    "    for i in range(x.shape[0] - 1):\n",
    "        y[q[i], q[i + 1]] += 1\n",
    "    y = y / y.sum(axis=1, keepdims=True)\n",
    "    y[np.isnan(y)] = 0\n",
    "    \n",
    "    for i in range(r.shape[0]):\n",
    "        for j in range(r.shape[1]):\n",
    "            r[i, j] = y[q[i], q[j]]\n",
    "    return r / 5. - 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/asif/anaconda3/envs/pytorch/lib/python3.7/site-packages/ipykernel_launcher.py:7: RuntimeWarning: invalid value encountered in true_divide\n",
      "  import sys\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[-0.9       , -0.9       , -0.95      , ..., -1.        ,\n",
       "        -1.        , -1.        ],\n",
       "       [-0.9       , -0.9       , -0.95      , ..., -1.        ,\n",
       "        -1.        , -1.        ],\n",
       "       [-1.        , -1.        , -1.        , ..., -1.        ,\n",
       "        -1.        , -1.        ],\n",
       "       ...,\n",
       "       [-1.        , -1.        , -1.        , ..., -0.80877193,\n",
       "        -0.80877193, -0.80877193],\n",
       "       [-1.        , -1.        , -1.        , ..., -0.80877193,\n",
       "        -0.80877193, -0.80877193],\n",
       "       [-1.        , -1.        , -1.        , ..., -0.80877193,\n",
       "        -0.80877193, -0.80877193]])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# MTF for 1 image\n",
    "get_mtf(x_train_3.values[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/asif/anaconda3/envs/pytorch/lib/python3.7/site-packages/ipykernel_launcher.py:7: RuntimeWarning: invalid value encountered in true_divide\n",
      "  import sys\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f74431339b0>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.imshow(get_mtf(x_train_3.values[0]), cmap='gray', origin='lower')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'divide': 'warn', 'over': 'warn', 'under': 'ignore', 'invalid': 'warn'}"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.seterr(divide='ignore', invalid='ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "smt3 = SMOTE(sampling_strategy={1: 30000, 2: 20000, 3: 20000, 4: 10000}, n_jobs=-1, random_state=42)\n",
    "x_train_3, y_train_3 = smt3.fit_sample(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "with h5py.File('/media/asif/data/hb_data_mit_bih_balanced_mtf.hdf5', mode='w') as hdf5_file:\n",
    "    hdf5_file.create_dataset(\"x_train\", (len(x_train_3), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_train\", (len(y_train_3),), np.int32)\n",
    "    hdf5_file.create_dataset(\"x_test\", (len(x_test), 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_test\", (len(y_test),), np.int32)\n",
    "    \n",
    "    transform = transforms.Compose([transforms.Resize((224, 224))])\n",
    "    \n",
    "    for i, (x, y) in enumerate(zip(x_train_3.values, y_train_3)):\n",
    "        mtf = get_mtf(x)\n",
    "        image = Image.fromarray(mtf.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_train\"][i, ...] = t\n",
    "        hdf5_file[\"y_train\"][i] = y\n",
    "\n",
    "    for i, (x, y) in enumerate(zip(x_test.values, y_test)):\n",
    "        mtf = get_mtf(x)\n",
    "        image = Image.fromarray(mtf.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        hdf5_file[\"x_test\"][i, ...] = t\n",
    "        hdf5_file[\"y_test\"][i] = y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 3 channels, this time with GAF, RP and MTF\n",
    "with h5py.File('/media/asif/data/hb_data_mit_bih_balanced_gaf_rp_mtf.hdf5', mode='w') as hdf5_file:\n",
    "    hdf5_file.create_dataset(\"x_train\", (len(x_train_3), 3, 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_train\", (len(y_train_3),), np.int32)\n",
    "    hdf5_file.create_dataset(\"x_test\", (len(x_test), 3, 224, 224), np.float32)\n",
    "    hdf5_file.create_dataset(\"y_test\", (len(y_test),), np.int32)\n",
    "    \n",
    "    transform = transforms.Compose([transforms.Resize((224, 224))])\n",
    "    \n",
    "    for i, (x1, x2, y) in enumerate(zip(x_gasf_train_3, x_train_3.values, y_train_3)):\n",
    "        image = Image.fromarray(x1.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        r = recurrence_plot(x2, steps=10)\n",
    "        mtf = get_mtf(x2)\n",
    "        image = Image.fromarray(r.astype(float))\n",
    "        t2 = np.array(transform(image))\n",
    "        image = Image.fromarray(mtf.astype(float))\n",
    "        t3 = np.array(transform(image))\n",
    "        hdf5_file[\"x_train\"][i, 0, ...] = t\n",
    "        hdf5_file[\"x_train\"][i, 1, ...] = t2\n",
    "        hdf5_file[\"x_train\"][i, 2, ...] = t3\n",
    "        hdf5_file[\"y_train\"][i] = y\n",
    "\n",
    "    for i, (x1, x2, y) in enumerate(zip(x_gasf_test_3, x_test.values, y_test)):\n",
    "        image = Image.fromarray(x1.astype(float))\n",
    "        t = np.array(transform(image))\n",
    "        r = recurrence_plot(x2, steps=10)\n",
    "        mtf = get_mtf(x2)\n",
    "        image = Image.fromarray(r.astype(float))\n",
    "        t2 = np.array(transform(image))\n",
    "        image = Image.fromarray(mtf.astype(float))\n",
    "        t3 = np.array(transform(image))\n",
    "        hdf5_file[\"x_test\"][i, 0, ...] = t\n",
    "        hdf5_file[\"x_test\"][i, 1, ...] = t2\n",
    "        hdf5_file[\"x_test\"][i, 2, ...] = t3\n",
    "        hdf5_file[\"y_test\"][i] = y"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}