irc_model_logreg.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Logistic Regression Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from irc_utils import compute_score, exponential_moving_average\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV\n",
    "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score, f1_score\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.linear_model import LinearRegression, LogisticRegression "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " #### Data processing\n",
    "- Each row in csv file represent one session\n",
    "- Each session is represented by\n",
    "    - periodicity, duration, total packet size, message count, number of source ports, destination port"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>periodicity</th>\n",
       "      <th>duration</th>\n",
       "      <th>pkt_size</th>\n",
       "      <th>msg_count</th>\n",
       "      <th>src_ports_count</th>\n",
       "      <th>dst_port</th>\n",
       "      <th>src_spec_chars</th>\n",
       "      <th>msg_spec_chars</th>\n",
       "      <th>msg_word_entropy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.0</td>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.000000</td>\n",
       "      <td>900.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.277307</td>\n",
       "      <td>108163.621111</td>\n",
       "      <td>16480.378889</td>\n",
       "      <td>56.635556</td>\n",
       "      <td>1.0</td>\n",
       "      <td>45321.533333</td>\n",
       "      <td>0.259800</td>\n",
       "      <td>0.254657</td>\n",
       "      <td>4.734015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.276155</td>\n",
       "      <td>129238.384475</td>\n",
       "      <td>44542.083275</td>\n",
       "      <td>156.658121</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10760.013916</td>\n",
       "      <td>0.135193</td>\n",
       "      <td>0.171088</td>\n",
       "      <td>3.027613</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.008200</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>112.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2441.000000</td>\n",
       "      <td>0.095200</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.069475</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>288.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>37789.000000</td>\n",
       "      <td>0.157900</td>\n",
       "      <td>0.177175</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.277200</td>\n",
       "      <td>12866.500000</td>\n",
       "      <td>1291.500000</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>45225.000000</td>\n",
       "      <td>0.230800</td>\n",
       "      <td>0.228350</td>\n",
       "      <td>4.889700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.277200</td>\n",
       "      <td>259333.250000</td>\n",
       "      <td>7961.500000</td>\n",
       "      <td>48.250000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>52808.000000</td>\n",
       "      <td>0.307700</td>\n",
       "      <td>0.288325</td>\n",
       "      <td>7.132750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>619200.000000</td>\n",
       "      <td>402599.000000</td>\n",
       "      <td>3592.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>65124.000000</td>\n",
       "      <td>0.846200</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>10.707400</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       periodicity       duration       pkt_size    msg_count  \\\n",
       "count   900.000000     900.000000     900.000000   900.000000   \n",
       "mean      0.277307  108163.621111   16480.378889    56.635556   \n",
       "std       0.276155  129238.384475   44542.083275   156.658121   \n",
       "min       0.008200       0.000000     112.000000     1.000000   \n",
       "25%       0.069475      16.000000     288.000000     2.000000   \n",
       "50%       0.277200   12866.500000    1291.500000     8.000000   \n",
       "75%       0.277200  259333.250000    7961.500000    48.250000   \n",
       "max       1.000000  619200.000000  402599.000000  3592.000000   \n",
       "\n",
       "       src_ports_count      dst_port  src_spec_chars  msg_spec_chars  \\\n",
       "count            900.0    900.000000      900.000000      900.000000   \n",
       "mean               1.0  45321.533333        0.259800        0.254657   \n",
       "std                0.0  10760.013916        0.135193        0.171088   \n",
       "min                1.0   2441.000000        0.095200        0.000000   \n",
       "25%                1.0  37789.000000        0.157900        0.177175   \n",
       "50%                1.0  45225.000000        0.230800        0.228350   \n",
       "75%                1.0  52808.000000        0.307700        0.288325   \n",
       "max                1.0  65124.000000        0.846200        1.000000   \n",
       "\n",
       "       msg_word_entropy  \n",
       "count        900.000000  \n",
       "mean           4.734015  \n",
       "std            3.027613  \n",
       "min           -0.000000  \n",
       "25%            2.000000  \n",
       "50%            4.889700  \n",
       "75%            7.132750  \n",
       "max           10.707400  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Select model version\n",
    "plt.rcParams[\"figure.figsize\"] = (10,7)\n",
    "\n",
    "# Import data\n",
    "data = pd.read_csv('input/trn_data_small.csv', delimiter=';')\n",
    "\n",
    "# Prepare data for training\n",
    "# Divide the data into attributes and labels\n",
    "X = data.iloc[:, 1:-1]\n",
    "y = data.iloc[:, -1].values\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.4, random_state=0)\n",
    "\n",
    "# showing examples of data\n",
    "X_train.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1501, 9)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sc = StandardScaler()\n",
    "X_train = sc.fit_transform(X_train)\n",
    "X_test = sc.transform(X_test)\n",
    "X = sc.transform(X)\n",
    "X.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Exhaustive Grid Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 40 candidates, totalling 200 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.\n",
      "[Parallel(n_jobs=-1)]: Done  34 tasks      | elapsed:    1.7s\n",
      "[Parallel(n_jobs=-1)]: Done 200 out of 200 | elapsed:    2.2s finished\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "param_grid = [\n",
    "    {'classifier' : [LogisticRegression()],\n",
    "     'classifier__penalty' : ['l1', 'l2'],\n",
    "    'classifier__C' : np.logspace(-4, 4, 20)\n",
    "    }\n",
    "]\n",
    "\n",
    "# Create grid search object\n",
    "pipe = Pipeline([('classifier' , LogisticRegression())])\n",
    "clf = GridSearchCV(pipe, param_grid = param_grid, cv = 5, verbose=True, n_jobs=-1)\n",
    "\n",
    "# Fit on data\n",
    "\n",
    "best_clf = clf.fit(X_train, y_train)\n",
    "logreg_model = best_clf.best_params_['classifier']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'classifier': LogisticRegression(C=1.623776739188721, class_weight=None, dual=False,\n",
       "                    fit_intercept=True, intercept_scaling=1, l1_ratio=None,\n",
       "                    max_iter=100, multi_class='warn', n_jobs=None, penalty='l1',\n",
       "                    random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                    warm_start=False),\n",
       " 'classifier__C': 1.623776739188721,\n",
       " 'classifier__penalty': 'l1'}"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_clf.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=1.623776739188721, class_weight=None, dual=False,\n",
       "                   fit_intercept=True, intercept_scaling=1, l1_ratio=None,\n",
       "                   max_iter=100, multi_class='warn', n_jobs=None, penalty='l1',\n",
       "                   random_state=None, solver='warn', tol=0.0001, verbose=0,\n",
       "                   warm_start=False)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "# C, kernel, gamma = clf.best_params_['C'], clf.best_params_['kernel'], clf.best_params_['gamma']\n",
    "logreg = logreg_model# LogisticRegression()\n",
    "\n",
    "logreg.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix:\n",
      "[[573   0]\n",
      " [  3  25]]\n",
      "\n",
      "Sensitivity(=Recall) TPR = TP / (TP + FN):\t\t0.8929\n",
      "Specificity SPC = TN / (FP + TN):\t\t\t1.0\n",
      "Precision PPV = TP / (TP + FP):\t\t\t\t1.0\n",
      "Negative Predictive Value NPV = TN / (TN + FN):\t\t0.9948\n",
      "False Positive Rate FPR = FP / (FP + TN)):\t\t0.0\n",
      "False Discovery rate FDR = FP / (FP + TP):\t\t0.0\n",
      "False Negative rate FNR = FN / (FN + TP):\t\t0.1071\n",
      "Accuraccy ACC = (TP + TN) / (P + N):\t\t\t0.995\n",
      "F1-score F1 = 2TP / (2TP + FP + FN):\t\t\t0.9434\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.8929, 1.0, 1.0, 0.9948, 0.0, 0.0, 0.1071, 0.995, 0.9434]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from irc_utils import compute_score\n",
    "\n",
    "y_pred = logreg.predict(X_test)\n",
    "\n",
    "compute_score(y_test, y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Improvement\n",
    "\n",
    "#### Higher number of samples?\n",
    "- could help in case we don't have overfitted model\n",
    "- To show the accuracy plot, Random Forrest classifier with tuned hyperparameters will be trained multiple times with different training dataset size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape: 80,9\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n",
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "train_ratio = np.arange(0.2,1,0.01)\n",
    "n_samples = len(X)\n",
    "train_size = list(map(lambda x: round(x*n_samples), train_ratio))\n",
    "\n",
    "score_trn_list = []\n",
    "score_tst_list = []\n",
    "for trn_ratio in train_ratio:\n",
    "    X_tmp, _, y_tmp, _ = train_test_split(X, y, train_size=trn_ratio, random_state=0)\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X_tmp, y_tmp, train_size=0.8, random_state=0)\n",
    "    logreg = LogisticRegression()\n",
    "    logreg.fit(X_train, y_train)\n",
    "    \n",
    "    y_pred_trn = logreg.predict(X_train)\n",
    "    y_pred_tst = logreg.predict(X_test)\n",
    "    \n",
    "    score_trn = compute_score(y_train, y_pred_trn, verbose=False)\n",
    "    score_tst = compute_score(y_test, y_pred_tst, verbose=False)\n",
    "    score_trn_list.append(score_trn)\n",
    "    score_tst_list.append(score_tst)\n",
    "\n",
    "print('shape: {},{}'.format(len(score_trn_list), len(score_trn_list[0])))\n",
    "score_trn_list = np.stack(score_trn_list, axis=0)\n",
    "score_tst_list = np.stack(score_tst_list, axis=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Sensitivity(=Recall) TPR = TP / (TP + FN):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_tpr = exponential_moving_average(score_trn_list[:,0], 0.1)\n",
    "tst_score_tpr = exponential_moving_average(score_tst_list[:,0], 0.1)\n",
    "\n",
    "plt.title(\"TPR per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('TPR')\n",
    "plt.plot(train_size, trn_score_tpr, label='Train TPR',color='navy')\n",
    "plt.plot(train_size, tst_score_tpr, label='Test TPR',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Specificity SPC = TN / (FP + TN):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_spc = exponential_moving_average(score_trn_list[:,1], 0.1)\n",
    "tst_score_spc = exponential_moving_average(score_tst_list[:,1], 0.1)\n",
    "\n",
    "\n",
    "plt.title(\"SPC per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('SPC')\n",
    "plt.plot(train_size, trn_score_spc, label='Train SPC',color='navy')\n",
    "plt.plot(train_size, tst_score_spc, label='Test SPC',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Precision PPV = TP / (TP + FP):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_ppv = exponential_moving_average(score_trn_list[:,2], 0.1)\n",
    "tst_score_ppv = exponential_moving_average(score_tst_list[:,2], 0.1)\n",
    "\n",
    "plt.title(\"PPV per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('PPV')\n",
    "plt.plot(train_size, trn_score_ppv, label='Train PPV',color='navy')\n",
    "plt.plot(train_size, tst_score_ppv, label='Test PPV',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Negative Predictive Value NPV = TN / (TN + FN):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_npv = exponential_moving_average(score_trn_list[:,3], 0.1)\n",
    "tst_score_npv = exponential_moving_average(score_tst_list[:,3], 0.1)\n",
    "\n",
    "plt.title(\"NPV per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('NPV')\n",
    "plt.plot(train_size, trn_score_npv, label='Train NPV',color='navy')\n",
    "plt.plot(train_size, tst_score_npv, label='Test NPV',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### False Positive Rate FPR = FP / (FP + TN)):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmEAAAG5CAYAAADGcOOUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3dfbRcdX3v8fd3Zs5JwmMgBKkEJSK9EkAiRJSqqJUqqBdYrYpcKBVR6lOtqFW4uhTxthf03oIWKkVEpUtBbr3a1KKoFCtXKxAwBSFSU0RJBAkhPObpnJnv/WPvOZmcnHCScCa/k+T9Wmuv/fTbe//27+yZ85m998yOzESSJElbV6N0BSRJknZEhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZNIRPwgIt5Wuh6S+s8QJulpi4h7I2JVRDzR0z0zIvaPiOyZdm9EnN2zXEbEk/W8pRHx1xHRLLkvkrS1GMIkTZT/mpm79HS/6Zk3PTN3AU4GPhYRx/bMO6ye93LgJOCt/a5oRLT6vQ1JGo8hTNJWk5n/BtwJHDLGvMXAj4C5G1u+PpN2TkTcFRErIuKLETG1Z/7rI2JhRDwSET+OiOePWvbDEXE78OToIBaVCyPiwYh4LCLuiIhD6nmvi4if1tPvi4hze5brnu07vZ63IiLeEREvjIjb67pc3FP+LRHxo4i4OCIejYifR8SrnmKf3xoRi+r1XhcRzx6vvpK2DYYwSVtFHRpeAhwM/HSM+c8DXgYsHmdVpwCvAQ4Afhf4aL38C4ArgD8FZgB/B8yPiCk9y54MvI7qzNzwqPW+Gji6XufuwJuA5fW8J4HTgOn18u+MiBNHLf8i4ECqs3kXAR8Bjqn3900R8fJRZf8T2Av4OPB/I2LPMdrkBOC/A38IzARuBK7ahPpK2gYYwiRNlG/WZ30eiYhvjpr3EPAwcDlwdmZe3zPvtoh4ElgE/AD423G2c3Fm3peZDwN/SRWsAM4E/i4zb8rMdmZ+GVgDvLhn2c/Wy64aY71DwK7A84DIzEWZeT9AZv4gM+/IzE5m3k4VhF4+avlPZubqzPwuVWi7KjMfzMylVOHpBT1lHwQuysyhzPwacDdVuBvtHcD/rOsyDPwVMLc+G7bR+kraNhjCJE2UEzNzet2NPku0V2bukZkHZeZnR807HNiF6gzSi4Cdx9nOfT3DvwKeWQ8/G/hATxB8BNivZ/7oZdeTmf8CXAxcAjwYEZdFxG4AEfGiiLghIpZFxKNU4WivUav4bc/wqjHGd+kZX5qZuZH96PVs4DM9+/MwEMC+T1VfSdsGQ5ik4rJyDfBvwMfGKb5fz/CzgO4XAO4D/rInCE7PzJ0y86qe8r3BZ6x6fDYzjwDmUF3m+4t61leB+cB+mbk7cClVGNpS+0ZE7/K9+9HrPuBPR+3TtMz88Tj1lbQNMIRJmkzOB94eEfs8RZl3R8Ss+h6qjwBfq6d/HnhHfdYqImLn+ob6XTdlw/WN9C+KiAGqy4mrgU49e1fg4cxcHRFHAv9tS3aux97AeyNiICLeCBwEXDtGuUuBcyLi4LqOu9flx6uvpG2AIUzSpJGZdwA/5KnP6HwV+C5wD9XN7f+jXnYB8HaqS3QrqG7wf8tmbH43qiC3gury4HLg0/W8dwHnRcTjVGfqrtmM9Y7lJqqb+B+iuq/tDZm5wU31mfkN4ALg6oh4DPgZcNwm1FfSNiDWvy1BkiaviLgXeFtmfr90XbZURLyFah9eWrouksryTJgkSVIBhjBJkqQCvBwpSZJUgGfCJEmSCtjmHmK711575f7771+6GpIkSeO69dZbH8rMmWPN2+ZC2P7778+CBQtKV0OSJGlcEfGrjc3zcqQkSVIBhjBJkqQCDGGSJEkFbHP3hEmSpIkzNDTEkiVLWL16demqbNOmTp3KrFmzGBgY2ORlDGGSJO3AlixZwq677sr+++9PRJSuzjYpM1m+fDlLlixh9uzZm7yclyMlSdqBrV69mhkzZhjAnoaIYMaMGZt9NtEQJknSDs4A9vRtSRsawiRJkgowhEmSpGKWL1/O3LlzmTt3Lvvssw/77rvvyPjatWs3aR2nn346d9999yZv8/LLL2fmzJkj2zn99NMBOPXUU5k9ezZz587liCOO4Kabbtpg+mGHHcYNN9yw+Ts6Bm/MlyRJxcyYMYOFCxcCcO6557LLLrvwwQ9+cL0ymUlm0miMfe7oi1/84mZv95RTTuGiiy7aYPqFF17IiSeeyLXXXss73/lObrvttvWmf+973+Nd73oXixYt2uxtjuaZMEmSNOksXryYOXPmcMopp3DwwQdz//33c+aZZzJv3jwOPvhgzjvvvJGyL33pS1m4cCHDw8NMnz6ds88+m8MOO4yjjjqKBx98cIu2f/TRR7N48eINph911FEsXbp0i/erl2fCJEkSAO9733dYuPCBCV3n3Ln7cNFFx27Rsj//+c+58sormTdvHgDnn38+e+65J8PDw7zyla/kDW94A3PmzFlvmUcffZSXv/zlnH/++bz//e/niiuu4Oyzz95g3V/5ylf4wQ9+AMD73/9+TjvttPXm/9M//ROHHnroBst95zvf4cQTT9yi/RnNECZJkialAw44YCSAAVx11VV84QtfYHh4mN/85jfcddddG4SwadOmcdxxxwFwxBFHcOONN4657o1djjzrrLM499xz2Xvvvfn85z+/3vQPfehDLF26dOResaerbyEsIq4AXg88mJmHjDE/gM8ArwVWAm/JzNv6VR9JkvTUtvSMVb/svPPOI8O/+MUv+MxnPsPNN9/M9OnTOfXUU8f8Xa7BwcGR4WazyfDw8GZts3vv18amX3jhhZxxxhkTEsT6eU/Yl4Cn+mseBxxYd2cCn+tjXSRJ0jbsscceY9ddd2W33Xbj/vvv57rrritSj/e9732sXLmS66+//mmvq28hLDN/CDz8FEVOAK7Myk+A6RHxO/2qjyRJ2nYdfvjhzJkzh+c973mcdtppvOQlLylSj4jgox/9KJ/61Kee/roycwKqtJGVR+wPfGsjlyO/BZyfmf+vHr8e+HBmLhij7JlUZ8t41rOedcSvfvWrvtVZkqQdyaJFizjooINKV2O7MFZbRsStmTlvrPLbxE9UZOZlmTkvM+fNnDmzdHUkSZKetpIhbCmwX8/4rHqaJEnSdq9kCJsPnBaVFwOPZub9BesjSZK01fTzJyquAl4B7BURS4CPAwMAmXkpcC3Vz1MspvqJitP7VRdJkqTJpm8hLDNPHmd+Au/u1/YlSZIms23ixnxJkqTtjY8tkiRJxSxfvpxXvepVADzwwAM0m026v4Rw8803r/cL+E/liiuu4LWvfS377LPPBvNOPfVUfvSjH7H77rsD8Pa3v513v/vdzJo1iz322IOI4JnPfCZXXnkle++993rTZ8yYwZVXXsl+++23wXqfLs+ESZKkYmbMmMHChQtZuHAh73jHOzjrrLNGxjc1gEEVwh54YOMPH7/wwgtH1vvud6+7G+rGG2/k9ttv5/nPfz7nn3/+BtN/7/d+j7/6q7/asp0bhyFMkiRNSl/+8pc58sgjmTt3Lu9617vodDoMDw/zx3/8xxx66KEccsghfPazn+VrX/saCxcu5KSTTmLu3LmsXbt2s7d19NFHs3jx4g2mH3XUUSxd2p9f0PJypCRJqtzwPnhw4cSuc++58MqLNnuxn/3sZ3zjG9/gxz/+Ma1WizPPPJOrr76aAw44gIceeog77rgDgEceeYTp06fzN3/zN1x88cXMnTt3zPWdddZZnHvuuQB89atfZc6cOSPzMpNvfetbHHrooRssd9111435QO+JYAiTJEmTzve//31uueUW5s2rnvizatUq9ttvP17zmtdw99138973vpfXve51vPrVr96k9V144YVjhqmXvexlNBoN5s6dy4c//OH1pi9fvpzp06evd5lyIhnCJElSZQvOWPVLZvLWt76VT37ykxvMu/322/n2t7/NJZdcwte//nUuu+yyLd7OjTfeyPTp08ecvvPOO3PyySfziU98YkIe2D2a94RJkqRJ55hjjuGaa67hoYceAqpvUf76179m2bJlZCZvfOMbOe+887jtttsA2HXXXXn88ccntA4DAwNcdNFFXHHFFTzyyCMTum7wTJgkSZqEDj30UD7+8Y9zzDHH0Ol0GBgY4NJLL6XZbHLGGWeQmUQEF1xwAQCnn346b3vb25g2bdpm/bTFeGbNmsUb3/hGPve5z3HOOedMyDq7ovrh+m3HvHnzcsGCBaWrIUnSdmHRokUcdNBBpauxXRirLSPi1sycN1Z5L0dKkiQVYAiTJEkqwBAmSdIOblu7NWky2pI2NIRJkrQDmzp1KsuXLzeIPQ2ZyfLly5k6depmLee3IyVJ2oHNmjWLJUuWsGzZstJV2aZNnTqVWbNmbdYyhjBJknZgAwMDzJ49u3Q1dkhejpQkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBXQ1xAWEcdGxN0RsTgizh5j/rMi4oaI+GlE3B4Rr+1nfSRJkiaLvoWwiGgClwDHAXOAkyNizqhiHwWuycwXAG8G/rZf9ZEkSZpM+nkm7EhgcWbek5lrgauBE0aVSWC3enh34Dd9rI8kSdKk0c8Qti9wX8/4knpar3OBUyNiCXAt8GdjrSgizoyIBRGxYNmyZf2oqyRJ0lZV+sb8k4EvZeYs4LXA30fEBnXKzMsyc15mzps5c+ZWr6QkSdJE62cIWwrs1zM+q57W6wzgGoDM/DdgKrBXH+skSZI0KfQzhN0CHBgRsyNikOrG+/mjyvwaeBVARBxEFcK83ihJkrZ7fQthmTkMvAe4DlhE9S3IOyPivIg4vi72AeDtEfHvwFXAWzIz+1UnSZKkyaLVz5Vn5rVUN9z3TvtYz/BdwEv6WQdJkqTJqPSN+ZIkSTskQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKqCvISwijo2IuyNicUScvZEyb4qIuyLizoj4aj/rI0mSNFm0+rXiiGgClwB/ACwBbomI+Zl5V0+ZA4FzgJdk5oqI2Ltf9ZEkSZpM+nkm7EhgcWbek5lrgauBE0aVeTtwSWauAMjMB/tYH0mSpEmjnyFsX+C+nvEl9bRevwv8bkT8KCJ+EhHHjrWiiDgzIhZExIJly5b1qbqSJElbT+kb81vAgcArgJOBz0fE9NGFMvOyzJyXmfNmzpy5lasoSZI08foZwpYC+/WMz6qn9VoCzM/Mocz8JfAfVKFMkiRpu9bPEHYLcGBEzI6IQeDNwPxRZb5JdRaMiNiL6vLkPX2skyRJ0qTQtxCWmcPAe4DrgEXANZl5Z0ScFxHH18WuA5ZHxF3ADcBfZObyftVJkiRpsojMLF2HzTJv3rxcsGBB6WpIkiSNKyJuzcx5Y80rfWO+JEnSDskQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIK2OIQFhHPmsiKSJIk7UjGDWERcVREvCEi9q7Hnx8RXwV+1PfaSZIkbaeeMoRFxKeBK4A/Av45Iv4H8F3gJuDA/ldPkiRp+9QaZ/7rgBdk5uqI2AO4DzgkM+/te80kSZK2Y+NdjlydmasBMnMF8AsDmCRJ0tM33pmw50TE/Ho4gNk942Tm8X2rmSRJ0nZsvBB2wqjx/9WvikiSJO1InjKEZea/RsRc4LnAnZm5aOtUS5Ikafs23rcjPwZcw7pvR759q9RKkiRpOzfe5ciTgLmZuTIiZgDfAT7f/2pJkiRt38b7duSazFwJkJnLN6G8JEmSNsHmfjvyAL8dKUmS9PT57UhJkqQCxgthv8zMX2+VmkiSJO1AxrvH65vdgYj4ep/rIkmStMMYL4RFz/Bz+lkRSZKkHcl4ISw3MixJkqSnYbx7wg6LiMeozohNq4epxzMzd+tr7SRJkrZT4z22qLm1KiJJkrQj8cdXJUmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgF9DWERcWxE3B0RiyPi7Kco90cRkRExr5/1kSRJmiz6FsIioglcAhwHzAFOjog5Y5TbFfhz4KZ+1UWSJGmy6eeZsCOBxZl5T2auBa4GThij3CeBC4DVfayLJEnSpNLPELYvcF/P+JJ62oiIOBzYLzP/+alWFBFnRsSCiFiwbNmyia+pJEnSVlbsxvyIaAB/DXxgvLKZeVlmzsvMeTNnzux/5SRJkvqsnyFsKbBfz/iselrXrsAhwA8i4l7gxcB8b86XJEk7gn6GsFuAAyNidkQMAm8G5ndnZuajmblXZu6fmfsDPwGOz8wFfayTJEnSpNC3EJaZw8B7gOuARcA1mXlnRJwXEcf3a7uSJEnbglY/V56Z1wLXjpr2sY2UfUU/6yJJkjSZ+Iv5kiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBfQ1hEXFsRNwdEYsj4uwx5r8/Iu6KiNsj4vqIeHY/6yNJkjRZ9C2ERUQTuAQ4DpgDnBwRc0YV+ykwLzOfD/wD8Kl+1UeSJGky6eeZsCOBxZl5T2auBa4GTugtkJk3ZObKevQnwKw+1keSJGnS6GcI2xe4r2d8ST1tY84Avj3WjIg4MyIWRMSCZcuWTWAVJUmSypgUN+ZHxKnAPODTY83PzMsyc15mzps5c+bWrZwkSVIftPq47qXAfj3js+pp64mIY4CPAC/PzDV9rI8kSdKk0c8zYbcAB0bE7IgYBN4MzO8tEBEvAP4OOD4zH+xjXSRJkiaVvoWwzBwG3gNcBywCrsnMOyPivIg4vi72aWAX4P9ExMKImL+R1UmSJG1X+nk5ksy8Frh21LSP9Qwf08/tS5IkTVaT4sZ8SZKkHY0hTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBrdIVmHSefAB+eysQEA2IAOp+NOquWXWN5rrh1lTYeR+YMr1eRpIkaeMMYaP95t9g/h9u+fLNKVUY2/l3erpnwE57w06j+oO7GtgkSdpBGcJG2+8VcMrNkB3IrPrU/ZGuXXWdnv7wyuos2pMPwJP3V92Ku2HJD2D1irG31ZwC0/aqu5k9w3vB1D1h2p5Vf+qeMHVG1Z+ye3UGTpIkbdMMYaNN3QP2eeHErrO9FlYug5UPwqoHq/6Tv636q5fDqoeq+Y//qhreWGjrGtytuuw5dTpM2WPd8OBuVUgb3B2m7Fb3d6/OuA3uVvfrrjk4sfsoSZI2iyFsa2gOwq77Vt2m6AzD6kdg9cM93fK6vwLWPFJ13eFH74EHH4G1j8Gax4DchDpNqcLYwC4wuAsM7FwND9TD2YHO0LquXfejsa5MtxvcBVrTqnq31264TGOgLrtT1W/ttG68tVO17EDd747TqdbVXV+3D3W5nq67bMPDWZK07ejrf62IOBb4DNAELs/M80fNnwJcCRwBLAdOysx7+1mnbUKjBTvtVXWbKzuw9ok6kD1adUOPw9rHq4DWHV77eFVm6MmqW/sEDD0Bq5ZV49GswlNjAJp1v9GqwtDq+2D4yfWXzfa6uneX6y7bHqrKD6+e2HYaLZp1MJsKzalVvzWtCpytqVV/9HBzShWSxxwfhMZgPT7YM97dv97h0W3VO95a134xxheSM+vAuWZdBz3raa1bVzS9j1CVzOpY6b4Oh1dVH4R6b5nITn3bRH0rBZ31b62AUcdqfUxHq3o/GPnAV/fXrKiO1d7XR2vK+q+TkddFT7/3tdEc3PA1Eq3132caA9vPcd79GzRa288+9UNm/X9qBax6GNqrNzxe1nt/bY3qNzdcH1n1u1+sm4T6FsIioglcAvwBsAS4JSLmZ+ZdPcXOAFZk5nMj4s3ABcBJ/arTZJWZDA93GBrqMDTUZmiow/Dw+t3QUHu98XY7e4bXn95ud/tJu70r7fbOdDrPGJnX6VTzqn413jttU7vMdf3IYWi0aDQaNJsNGo2g2Yy6X403GsmUxhqmNtcypbGawVhDo7OaZq6mlatpsqbq52qIJp0YIGOATlQvuk4MEJE0cw3NrMq2WF0Pr6HJWppD64ZbrKHF2mo4nqDJCloMVdNiXb/JMK0YosUQjej09W/dyQbtbNChSScbtGKIgcbwZq2jnU3a2aRDs1pXNmnTopONav006WSTDt15TTIbdKi2nd3laJB1ucyoy1fTOzRIGiPLdacnUa2zE3QyaHcadT9oZzWt0+kZ7wSdhKQql/U3jTtZrSvrdUKDTj0ODYIOO7VWsXPzSXZurWSn1kp2bj7JTq2VDDbW0oo2zWjTiqGq3ximQWdkf0f2o96/Bh1aMUwz2jTrfiuGgRxZZqQtu+2YTZLGSFt36r9b0KmXX7euZrRpRLtn+eZIe3eyQZAj5UZ3jWjTpEMj2jSiQ5Puuhr137W7rqoeDdpMaaypXj99Pl5H62TQzhbNGKYRm3DG/WlY9zppjhzT3X5Sz+s51rP+2zSiQ3SP4u5wdGjQJkaOsg4R1fC6day/XugelevKNqg+bK7bdmO9OlTvJdX7zkCspcVaWjG04T7RWm+7vV1u0G/0jFf72Ymqn/T062md3mkEGb3l1o3TfY1Ho35dVq+Z7nSitx8jw+vapTpeI6vjOLIep01kt73b9d+iXb0/Z/2enGtoUr2PT8nHGczHGOw8NtK+W6J6H4HYyJWgah9adKJqy4wWnWjx2JwPsferP7TF2326+nkm7EhgcWbeAxARVwMnAL0h7ATg3Hr4H4CLIyIys7+v7qfwrW/9Byef/PWtsq12uzMSuCazCOoQFURU4SqiO17Ng/qD+Rghb/P+mi1gl57xBNbW3WhT6m7T9yMinrLfanaY2moz2Oow0GwzpdlmSqvDQHOYKa02A802g80Og602A402A83OyPRWo8NAs1NP7463aUWHVrNDq1FN605vRLK23Rrp1rSbrBkeYKjdhICBkfLVdppRLdeMalqz0WGg0aHZaNf9qkyz3sa64TbNSJqNdhU6G52e8aQRWf3zr6dXw3V/1Hgjkma331i3bCOSVqM/x/GTQ4M8sXYaj62ZxuMrp/L48CBDnQGG2lMZ7jRZ224y3GnSzhipX7XP1T/fVqNDuxMMdZoMtZsMdZoMtxsMdZpkBs1mh4FGjrRbt+suW7VDh2ZjmAav/YMAAAtuSURBVGasIbNa13BngKH2lHq4CqbNUetpNpJWdGhnMNyZynCnwXCnwVC7Cmvd4eFOg3YnRua3O426TYfX/R2jOh7aGawansLKoUFWDQ2yqj2FVUODrGkPMFQv2w3Gw506FI8E4gbtThWmOhmQyUCrUx3Tzeo4G2xV23ly7SCPrp7GI6un1v1pPLFmoAqVkbQaydSBNoPNNoOtNlNa9XCzw2BruFpXPa/VDa2NNgON4Xpf2vXfad28VnTWTW906qDae9yv/7dtjvp7dbJVfQCo27PdaTBct2u3TTpZtUu70yCT6m/UXc/I66x636pC5/rLAvXrp7Pecq1Gh7Xtqawa2oPVQy1WD7dYPTzA6uEWnYyR+neX675/dF933XV03x/WbWN4pI7NnuOqGaP69XoGe+Z3X6fNRve1m/W09V/DzcbE/MvtdKqgWX0IazDcbtTHfoNVQwOsHmrx+HCLVd32GWrxyKpdeHjVTFasnMbDq6bx8MpprFg5ldXDrZH30+r9tmqXgea6NlpvuH7/SSAz1usHrGv3Ue2859oWb3r1hOz+FulnCNsXuK9nfAnwoo2VyczhiHgUmAE81FsoIs4EzqxHn4iIu/tS423TXoxqr4lWhauq25ZlfWZ6PI9Xvb636w5oC9u0G8IfneDqbDc8VieebbrFuh/GxvyHMfna9eu3ctKHP9DvrTx7YzO2iTuZM/My4LLS9ZiMImJBZs4rXY/tje068WzT/rBdJ55t2h+264b6eafaUmC/nvFZ9bQxy0REC9id6gZ9SZKk7Vo/Q9gtwIERMTsiBoE3A/NHlZkP/Ek9/AbgX0reDyZJkrS19O1yZH2P13uA66h+ouKKzLwzIs4DFmTmfOALwN9HxGLgYaqgps3jZdr+sF0nnm3aH7brxLNN+8N2HSU88SRJkrT1Tc5fL5MkSdrOGcIkSZIKMIRtAyKiGRE/jYhv1eOzI+KmiFgcEV+rv/hAREypxxfX8/cvWe/JLCKmR8Q/RMTPI2JRRBwVEXtGxPci4hd1f4+6bETEZ+t2vT0iDi9d/8kqIs6KiDsj4mcRcVVETPV43TwRcUVEPBgRP+uZttnHZkT8SV3+FxHxJ2Nta0eykXb9dP0ecHtEfCMipvfMO6du17sj4jU904+tpy2OiLO39n5MJmO1ac+8D0RERsRe9bjH6hgMYduGPwcW9YxfAFyYmc8FVlA9/gl6HgMFXFiX09g+A3wnM58HHEbVvmcD12fmgcD19TjAccCBdXcm8LmtX93JLyL2Bd4LzMvMQ6i+kNN9HJnH66b7EnDsqGmbdWxGxJ7Ax6l+IPtI4OPd4LYD+xIbtuv3gEMy8/nAfwDnAETEHKpj9+B6mb+tPwx3H8d3HDAHOLkuu6P6Ehu2KRGxH/Bq4Nc9kz1Wx2AIm+QiYhbwOuDyejyA36d6zBPAl4ET6+ET6nHq+a+qy6tHROwOHE317Vwyc21mPsL67Te6Xa/Myk+A6RHxO1u52tuKFjCt/t2/nYD78XjdLJn5Q6pvi/fa3GPzNcD3MvPhzFxBFTY2+Ge5IxmrXTPzu5nZfXjrT6h+zxKqdr06M9dk5i+BxVQBYeRxfJm5Fug+jm+HtJFjFaoPVR+C9R7k6LE6BkPY5HcR1cHcfRbEDOCRnjeOJVSPf4JRj4Gies7LjK1X1W3GbGAZ8MX6Mu/lEbEz8IzMvL8u8wDwjHp4rEdw7YvWk5lLgf9F9en3fqrj71Y8XifC5h6bHrOb763At+th23ULRcQJwNLM/PdRs2zTMRjCJrGIeD3wYGbeWrou25kWcDjwucx8AfAk6y7vAFD/aLC/37IZ6ksIJ1CF3GcCO7MDfaLdWjw2J15EfAQYBr5Sui7bsojYCfjvwMdK12VbYQib3F4CHB8R91Kd9v59qnuZpteXe2D9x0H5GKhNswRYkpk31eP/QBXKftu9zFj3H6znb8ojuATHAL/MzGWZOQT8X6pj2OP16dvcY9NjdhNFxFuA1wOn9DyxxXbdMgdQfQj79/r/1izgtojYB9t0TIawSSwzz8nMWZm5P9VNov+SmacAN1A95gmqxz79Yz3sY6A2QWY+ANwXEf+lnvQq4C7Wb7/R7Xpa/e2eFwOP9lwa0jq/Bl4cETvV93Z129Xj9enb3GPzOuDVEbFHfYby1fU09YiIY6lu9zg+M1f2zJoPvLn+Bu9sqpvJb2bTHse3w8rMOzJz78zcv/6/tQQ4vH7P9VgdS2babQMd8ArgW/Xwc6jeEBYD/weYUk+fWo8vruc/p3S9J2sHzAUWALcD3wT2oLof6XrgF8D3gT3rskH1jaj/BO6g+vZf8X2YjB3wCeDnwM+AvwemeLxudhteRXVP3RDVP7EztuTYpLrHaXHdnV56v0p3G2nXxVT3Iy2su0t7yn+kbte7geN6pr+W6puU/wl8pPR+TbY2HTX/XmCvethjdYzOxxZJkiQV4OVISZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJmlCRERGxP/uGf9gRJw7Qev+UkS8YfyST3s7b4yIRRFxQ7+3NU497o2IvUrWQVL/GcIkTZQ1wB9OtvDQ82v9m+IM4O2Z+cp+1UeSugxhkibKMHAZcNboGaPPZEXEE3X/FRHxrxHxjxFxT0ScHxGnRMTNEXFHRBzQs5pjImJBRPxH/VxVIqIZEZ+OiFsi4vaI+NOe9d4YEfOpfrV/dH1Ortf/s4i4oJ72MeClwBci4tOjyv9ORPwwIhbWy7ysnv65uk53RsQnesrfGxH/sy6/ICIOj4jrIuI/I+IdPXX8YUT8c0TcHRGXRsQG78kRcWrdHgsj4u/qfW7Wbfqzej82aHNJk9/mfEKUpPFcAtweEZ/ajGUOAw4CHgbuAS7PzCMj4s+BPwPeV5fbHziS6vl0N0TEc4HTqB5/8sKImAL8KCK+W5c/HDgkM3/Zu7GIeCZwAXAEsAL4bkScmJnnRcTvAx/MzAWj6vjfgOsy8y8jognsVE//SGY+XE+7PiKen5m31/N+nZlzI+JC4EtUz9GcSvU0gUvrMkcCc4BfAd8B/pDqWabduh4EnAS8JDOHIuJvgVOAO4F9M/OQutz0TWloSZOLZ8IkTZjMfAy4EnjvZix2S2ben5lrqB5p0g1Rd1AFr65rMrOTmb+gCmvPo3rO3GkRsRC4ierxPgfW5W8eHcBqLwR+kNWDxoeBrwBHj1dH4PT6HrdDM/PxevqbIuI24KfAwVSBqqv7TME7gJsy8/HMXAas6QlNN2fmPZnZpnoEzEtHbfdVVGHxlnofX0X1GKh7gOdExN/Uzz98bJz6S5qEPBMmaaJdBNwGfLFn2jD1h776kttgz7w1PcOdnvEO679HjX7GWlI9j+7PMnO9B/5GxCuAJ7es+hvKzB9GxNHA64AvRcRfAzcCHwRemJkrIuJLVGe6unr3Y/Q+dvdrrH3qFcCXM/Oc0XWKiMOA1wDvAN5E9fw9SdsQz4RJmlCZ+TBwDdVN7l33Up3RATgeGNiCVb8xIhr1fWLPoXqw8nXAOyNiACAifjcidh5nPTcDL4+IverLiCcD//pUC0TEs4HfZubngcupLnXuRhX0Ho2IZwDHbcE+HRkRs+tgehLw/0bNvx54Q0TsXddjz4h4dv3lh0Zmfh34aF0fSdsYz4RJ6of/DbynZ/zzwD9GxL9T3fu0JWepfk0VoHYD3pGZqyPicqpLlrdFRADLgBOfaiWZeX9EnA3cQHWm6Z8z8x/H2fYrgL+IiCHgCeC0zPxlRPwU+DlwH/CjLdinW4CLgefW9fnGqLreFREfpbpvrQEMAe8GVgFf7LmRf4MzZZImv8gcffZbktRv9SXTD2bm60vXRVIZXo6UJEkqwDNhkiRJBXgmTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgr4/8tByWth+eMMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_fpr = exponential_moving_average(score_trn_list[:,4], 0.1)\n",
    "tst_score_fpr = exponential_moving_average(score_tst_list[:,4], 0.1)\n",
    "\n",
    "plt.title(\"FPR per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('FPR')\n",
    "plt.plot(train_size, trn_score_fpr, label='Train FPR',color='navy')\n",
    "plt.plot(train_size, tst_score_fpr, label='Test FPR',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### False Discovery rate FDR = FP / (FP + TP):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_fdr = exponential_moving_average(score_trn_list[:,5], 0.1)\n",
    "tst_score_fdr = exponential_moving_average(score_tst_list[:,5], 0.1)\n",
    "\n",
    "plt.title(\"FDR per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('FDR')\n",
    "plt.plot(train_size, trn_score_fdr, label='Train FDR',color='navy')\n",
    "plt.plot(train_size, tst_score_fdr, label='Test FDR',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### False Negative rate FNR = FN / (FN + TP):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_fnr = exponential_moving_average(score_trn_list[:,6], 0.1)\n",
    "tst_score_fnr = exponential_moving_average(score_tst_list[:,6], 0.1)\n",
    "\n",
    "plt.title(\"FNR per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('FNR')\n",
    "plt.plot(train_size, trn_score_fnr, label='Train FNR',color='navy')\n",
    "plt.plot(train_size, tst_score_fnr, label='Test FNR',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Accuraccy ACC = (TP + TN) / (P + N): "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmEAAAG5CAYAAADGcOOUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3de5xddX3v/9dn75nJJBBAIBAlSChQIYLGENN6OV5RwfoDjvVG2x8WqtTTqtWW+pPqT5G2Hi+n3gqth1bAWhukeFQeVYuieOkFIWhECSIB0hDlEhIugWRue3/OH2vNZGcyGXKZne8wvJ6Px3qs617rs1ZW9n7v71qzV2QmkiRJ2rsapQuQJEl6IjKESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmahiJiTUScVLoOSd1jCJM0JSLiOxHxQETMmmDesoj4WkQ8GBEbI+L6iDirY/5+EfGJiFgbEY9ExO31+MF7dy8kae8xhEnaYxGxEPhvQAKnjpv3HODbwHeBo4GDgP8BnFLP7wO+BTwdOBnYD3gOsAFY1uW6e7q5fkmajCFM0lQ4E7gOuAx447h5HwU+m5kfzsz7s3JjZr6u47VPBf57Zq7KzHZm3peZf56ZX5toYxGREfH2iLgjIu6PiI9GRKNj/tkRcUvdMnd1RBwx7rV/GBG3AbdNsO7+iPjHiNhQt9zdEBGH1vPOqte7qd7273e87kURsS4i3hUR90XE3RFxekS8MiJ+XrcA/lnH8udHxJUR8YV6fT+MiGfuYH8bEfHuuoVwQ0RcEREHPla9kqY3Q5ikqXAm8Pm6e0VHaJlD1ap15SSvPQn418x8ZBe3+d+BpcAS4DTg7HqbpwF/BrwamAd8H1g+7rWnA78GLJpgvW8E9gcOp2q1ewuwpZ53H/Aqqta6s4CPR8SSjtfOB/qBw4D3AX8H/A5wIlVL4f8fEUd2LH8a8M/AgcA/AV+OiN4JanpbXfMLgacADwAX7US9kqYxQ5ikPRIRzweOAK7IzBuB24Hfqmc/iep95u5JVnHQY8zfkQ9n5sbMXAt8Ajijnv4W4H9m5i2ZOQJ8EFjc2RpWz9+YmROFleG6pqMzs1W32j0MkJlfzczb69a87wLfoApXna/9y8wcBi4HDgY+mZmbMvNmYBXQ2dp1Y2ZeWS//MaoA9+sT1PQW4D2ZuS4zB4HzgdfUl1N3WK+k6c0QJmlPvRH4RmbeX4//E1svST4AtIEnT/L6DY8xf0fu6hj+L6oWIqgC4SfrS3MPAhuBoGqdmui1430OuBq4PCJ+GREfGW2diohTIuK6+tLig8ArqYLW2L5kZqseHg1493bM3wLsO1EdmdkG1nXsR6cjgC917NMtQAs4dLJ6JU1vhjBJuy0iZgOvA14YEfdExD3AO4FnRsQzM3Mz8J/Ab06ymmuoLmHus4ubP7xj+KnAL+vhu4Dfz8wDOrrZmfkfHcvnjlaamcOZ+YHMXAQ8l+ry45n1X31+EfhfwKGZeQDwNaqAt7vG9qG+p21Bx350ugs4Zdw+9WfmL3ZU7x7UJGkvMYRJ2hOnU7XILAIW191xVPdhjQaBdwG/GxF/GhEHAUTEMyPi8nr+56hCxhcj4tj6JvSDIuLPIuKVk2z7TyPiSRFxOPBHwBfq6Z8GzouIp9fb2j8iXruzOxQRL46IEyKiCTxMdbmvDfQBs4D1wEhEnAK8fGfXuwMnRsSr68uK7wAGqf7AYbxPA385ekk1IubV975NVq+kac4QJmlPvBG4NDPXZuY9ox1wIfDbEdFTt0C9pO7uiIiNwMVUrUjU9zidBPwM+CZVkLie6jLfDybZ9leAG4GVwFeBz9Tr+xLwYarLcw8DP6X+OYydNJ/qDwkeprrs913gc5m5CXg7cAXVZdbfAq7ahfXuaB9eX6/v/wVeXd8fNt4n6219IyI2UQW1X5us3j2sS9JeEJk7bJWXpGkpIhI4JjNXl65ld0XE+VQ30/9O6VoklWFLmCRJUgGGMEmSpAK8HClJklSALWGSJEkFPO4eXnvwwQfnwoULS5chSZL0mG688cb7M3PeRPMedyFs4cKFrFixonQZkiRJjyki/mtH87wcKUmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFdC2ERcQlEXFfRPx0B/MjIj4VEasj4qaIWNKtWiRJkqabbraEXQacPMn8U4Bj6u4c4G+7WIskSdK00rUQlpnfAzZOsshpwD9k5TrggIh4crfqkSRJmk56Cm77MOCujvF19bS7xy8YEedQtZbx1Kc+da8Up5kpM2m3k1YryUwAIoIIthmOekLn8J6qttum1cqx4dFaxo+P11nCjurZUZm5/erq6TuYUb9mfK2jw5NpNGLCrnN7o5vNdpvMpNVmbL93dEw6p+2o7p3Zz4mWGa0xonM4Ora7bQ2TrW90HRHbrnP8PnSuc6J1bXesxm2o8zwd3c5ozTs6ZhOtf/y6OusfXW58DZk7Hp6s9s7xKV22nTRoEQzX04N2JhC0M6pzORvV9Pq4dHaTHf+J5gEELXobbXpimGajRU+M0Iw2GU3a9NKKXtrRR0YPEY36OENE0hPD9ESbBiM0Y6TjOEJ2DEOMHd92RkcdSZMRmtGi2WjRZJieaNGMERrRpskIDepxWjRpEbRoRqveZpsGLRrRqvqj3dh4Nb9NMJK9DLd7Gckehtu9ddfDSFbTWlT9keyhlT0kTcZ2FkiqcymJ6rjQQzt6aGcP7eilRS+tbG5zLo3/t5+oX53TbPdvOdHrxxv//n766cfyspcdtf2Ce0nJELbTMvNi4GKApUuXTv4psIe+9a07ePvb/5WhodaE3Y4+hB7rg2H8GwhUJ0BPT4Pe3iY9PY16uDE23Gw2aDZjbLjqV2+6o/NGhzvf6Md3nR8u47vxHxaj0yZ6g99Rf9TQUIvBwREGB7ft7+jDdfRYdL7xT/Rh0vn6zteMD08ThZzxH3qT5I6dtqNAFJH094wwp2+YOb1Vf3Zv9UY43IJ2Bq12g3ZWHw4j7QaPDPbxyFAfm4d6gR2HvUa0md07XK+7Wu/sCfr9PSNV17ttf1ZPq+5vO9zXbNHX02JW3e9rVl1Poz1WY6sdtLKxdbjdoFXvy+i8VruqfVa9jtF1j453rnt0uLfZHtu/4Va1jmobDYZaTTZsns39j84Z69Y/sg/3b57DwHD1tjX6T5kZJNCIZFazNcG+jxCMfsDFNv3R14wel9F+X7M1ttzov1cmkMHYhyNBu711fdX+j9Bb73fnuoZbDbaM9LJluIctw70MjfZbzY73iBjbr4Dq36VnZLv62hkMDPcwMFJ1m+t1DYz0MDjSZHCkh8FW3a/H2xlEJFGfp6NnWrPR3ub8GD1mfc0WW4Z72DQ4i4cHZ7FpsI+HB2axaXAWA8M9NBpJI6ou6BgO6n61jdHh3mZ7+3Nz3HDnMZvVHNnmXBp/XvU221vP3Z7WuP+gE/9XGm41GMwmg1kfn1YPw63GdusaPU8bjeofpt3e+u+SGdW+Nnb+jWRguEkS9DVbNHfhdXvdjkobvV7W7M5mB0eaPDQwi4cG+rfpD4z00Aw6zrWt51M7G7SzQauj38rm2HBm0MoYWw6grzmyTVf93xphaNUb4WXndWfndkLJEPYL4PCO8QX1tKLmzp3FcccdTF9fk1mzeujra9DX16Svr0lvb5Nmc8cflI/VQtGZvqFK8SMjbUZG2gwPtzuGW7Ra1bzRADI6b6IWg1arzfBwjoWxnp7GNt/uR7fV2Y2uu/q2kPTFIHN6NjOruYX+5kD17bEdjLQ6PojbQTNazOkdYHbPAHN6Bti3b5DZPQPMao6wpT2Hzc392DJ7LgOz5zIY+zEY+5HN2TQa24fGRiMm/HaT2a6X73xN9bqIib4xtelhiJ4mNBvQaDboadLxuqDRgJ5Gg0aDjvE2vbmFPh6hj830soU+HqWPzfSxpR7fQm8MVNNiC30M0BsD9Nb9vrHhQfpicLfPuzYNhpnNSMxhOOYQJD05QJMBenKAHoZ2e90tmrTpoxV9tKi76Ku+hcZs2vTW3957quXoJWnSG21mRbv+hry1H7SJbFV9WkRW/WpbfWMtAS1Gv+n2sCX6eJRe2rG1S3ogqg+1rd/O29U3fIbZr/0Q89oP0tfaSN/IL+htPUAjh3f+mMYs2o1ZtKOv/k9XNTHUsanuGrSjj3ajr+pHH+2YQzt668/yhMh6X0f7bH19VtOop7VjH7JRHYOxddJbtUDkAI32AM0cpNEeoNHeRIzuz2ir7NZoSTZ6ycYsstFHNur1NvqAJFqDNNqDRGsL0d5MtAdptAaqaTlE5LhgshMymmSjn2zOIqOXaA/QGN7UsX9Tq9qvWdDsI5vVdmn0kc3Z0JgFzVlksw8afVuHm30QvdBTH5dmHwON3mqZRm/9ftdxLLP+92uPEDnM7PYgc9pDRGuQaA9Ba7ha59h2OjoaQNYZZKx5qv7WWG+zubWf0SRbI9AaIluDMDIArSF6W4NktslGH8ONXrKuNRu9dUtZ9aa29eMjO86D+gtnx3GLgIxestEzVkc2esjordYbzbrfC/Uy1b9tPS966mXqlqtGL+1oQjSr/5ONHpJGFXwYopFDNHOYRg7RyEEa7SGiPUS0h6E9BK0haA1WXY7+vxrdldE36RbZqpbN9nD9mmGyNUjv0CbmDT3EvKGHiaGHYLDuWlsgmhCNbTuAbEN7BLJV9dstyNF+u5qerXq8/r/QnLW16+mvh+fCkgVTdk7vjpIh7CrgrRFxOfBrwEOZud2lyL1t2bLDuPLK15UuY3vZhoEHYfABGKi7zuHhR+oPmhh3wgaMbK5eO/QQDD5Yn+R1f2gTDD1crb9bGr0wa/+q69t/63BzVrXt0f90o8NDm6BnFvTOhb650Ldf3Z9b7dM2yz4Egw9v/Y821Zp90Lsv9O7T0X9S3Z9T9XvmVF3vPtAzu5re2Y+ejjeG9tbh9kj17zb0CI3hTcwa2sSsoUeq/W8063XO2dofXf+kXT80+6t+PdxsNGkCvd05QntPZvXv3hpi7I1+mzf92Lr/zT4aEU/c3+Bpt7Z+MI59ONLxHtHxXlF/IEWjZ/sGpEwY2VId96FNVTcyUJ2fo+8vne810dh23aPbavZtPS+b/dAzi4jGJG2/jz87aIB7nNtnytYU4/qqdC2ERcRy4EXAwRGxDng/9edAZn4a+BrwSmA1sBk4q1u1TEtDj8Dme2HL/bB5PWxZv7W/5f662wADG6r+4AOTB6XObwjbzWvCrAO2hp9ZB8ABR1fhZtb+dcjZD2bV/d5963WN+zaRrSpQjAaizq7ZXwWigQdgYOPWbsvGOig91BGaHoIHb68+HEZrmHNoR1CbCyODMFy/6Q8+XA1vvreqZdb+sN8RW2uftX8V2Bqj7eVjTY8TjMfW6dHs2IeOoDca/nr3qb7panqIqP6t9dgaTWjU4X1PRNRfAObAPvOnpjZJY7oWwjLzjMeYn8Afdmv7u+3Re+CeFTBnHsw5pOp6d/LbQLtVhahHfll1j47276kCxKP3wuZ7qv7I5onX0dMPs+fB7IOh/0CYezjMPqjq+g+qpvU/CWY9qe4fUPV7Zm8NF6OtA6OtLo3eHd+1PZX6D6jCkSRJekyPixvz96pf/gdc9ZvbTuuZszWUNfqgPVxfh+7oDz8Km++b4LJYVAFqn/lVa8+TnwP7HApz5tchb14Vukb7vfvseWDa5nKAJEmajgxh4z31pfBbP6gvD95XXSLcfB9sqYfbw9DYt74hsmfsxkd6ZsM+T4Z9nwL7PKXq7/uUKnh5SUuSJI1jCBtv1v7w5GWlq5AkSTOc16skSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqYCuhrCIODkibo2I1RHx7gnmPzUiro2IH0XETRHxym7WI0mSNF10LYRFRBO4CDgFWAScERGLxi32XuCKzHwW8Abgb7pVjyRJ0nTSzZawZcDqzLwjM4eAy4HTxi2TwH718P7AL7tYjyRJ0rTRzRB2GHBXx/i6elqn84HfiYh1wNeAt020oog4JyJWRMSK9evXd6NWSZKkvar0jflnAJdl5gLglcDnImK7mjLz4sxcmplL582bt9eLlCRJmmrdDGG/AA7vGF9QT+v0e8AVAJn5n0A/cHAXa5IkSZoWuhnCbgCOiYgjI6KP6sb7q8YtsxZ4KUBEHEcVwrzeKEmSZryuhbDMHAHeClwN3EL1V5A3R8QFEXFqvdifAG+OiB8Dy4HfzczsVk2SJEnTRU83V56ZX6O64b5z2vs6hlcBz+tmDZIkSdNR6RvzJUmSnpAMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpgK6GsIg4OSJujYjVEfHuHSzzuohYFRE3R8Q/dbMeSZKk6aKnWyuOiCZwEfAyYB1wQ0RclZmrOpY5BjgPeF5mPhARh3SrHkmSpOmkmy1hy4DVmXlHZg4BlwOnjVvmzcBFmfkAQGbe18V6JEmSpo1uhrDDgLs6xtfV0zr9KvCrEfHvEXFdRJw80Yoi4pyIWBERK9avX9+lciVJkvae0jfm9wDHAC8CzgD+LiIOGL9QZl6cmUszc+m8efP2comSJElTr5sh7BfA4R3jC+ppndYBV2XmcGbeCfycKpRJkiTNaN0MYTcAx0TEkRHRB7wBuGrcMl+magUjIg6mujx5RxdrkiRJmha6FsIycwR4K3A1cAtwRWbeHBEXRMSp9WJXAxsiYhVwLfCnmbmhWzVJkiRNF5GZpWvYJUuXLs0VK1aULkOSJOkxRcSNmbl0onmlb8yXJEl6QjKESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBOwxhEdEfEfMmmD4vIvq7W5YkSdLMNllL2KeA/zbB9OcDH+9OOZIkSU8Mk4WwEzPz/4yfmJlfAl7QvZIkSZJmvslC2JzdfJ0kSZIew2Rh6r6IWDZ+YkQ8G1jfvZIkSZJmvp5J5v0pcEVEXAbcWE9bCpwJvKHLdUmSJM1oO2wJy8zrgWVAAL9bdwH8Wmb+YG8UJ0mSNFPtsCWs/nmKgzPz/eOmL4qIzEwvSUqSJO2mye4J+2vg4AmmHwR8sjvlSJIkPTFMFsKOzszvjZ+Ymd8HntG9kiRJkma+yULY3Enm9U51IZIkSU8kk4Ww1RHxyvETI+IU4I7ulSRJkjTzTfYTFe8AvhoRr2Pbn6h4DvCqbhcmSZI0k032ExW3AScA3wUWAkcA3wHeBPzRXqhNkiRpxpqsJYzMHAQujYglwBnA+4E7gS/uhdokSZJmrMl+J+xXqYLXGcD9wBeAyMwX76XaJEmSZqzJWsJ+BnwfeFVmrgaIiHfulaokSZJmuMn+OvLVwN3AtRHxdxHxUqrHFkmSJGkPTXZj/pcz8w3AscC1VH8teUhE/G1EvHxvFShJkjQTTdYSBkBmPpqZ/5SZ/w+wAPgR8P91vTJJkqQZ7DFDWKfMfCAzL87Ml3arIEmSpCeCXQphkiRJmhqGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBXQ1RAWESdHxK0RsToi3j3Jcr8ZERkRS7tZjyRJ0nTRtRAWEU3gIuAUYBFwRkQsmmC5ucAfAT/oVi2SJEnTTTdbwpYBqzPzjswcAi4HTptguT8HPgwMdLEWSZKkaaWbIeww4K6O8XX1tDERsQQ4PDO/OtmKIuKciFgRESvWr18/9ZVKkiTtZcVuzI+IBvAx4E8ea9nMvDgzl2bm0nnz5nW/OEmSpC7rZgj7BXB4x/iCetqoucDxwHciYg3w68BV3pwvSZKeCLoZwm4AjomIIyOiD3gDcNXozMx8KDMPzsyFmbkQuA44NTNXdLEmSZKkaaFrISwzR4C3AlcDtwBXZObNEXFBRJzare1KkiQ9HvR0c+WZ+TXga+OmvW8Hy76om7VIkiRNJ/5iviRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUQFdDWEScHBG3RsTqiHj3BPP/OCJWRcRNEfGtiDiim/VIkiRNF10LYRHRBC4CTgEWAWdExKJxi/0IWJqZzwCuBD7SrXokSZKmk262hC0DVmfmHZk5BFwOnNa5QGZem5mb69HrgAVdrEeSJGna6GYIOwy4q2N8XT1tR34P+PpEMyLinIhYEREr1q9fP4UlSpIklTEtbsyPiN8BlgIfnWh+Zl6cmUszc+m8efP2bnGSJEld0NPFdf8COLxjfEE9bRsRcRLwHuCFmTnYxXokSZKmjW62hN0AHBMRR0ZEH/AG4KrOBSLiWcD/Bk7NzPu6WIskSdK00rUQlpkjwFuBq4FbgCsy8+aIuCAiTq0X+yiwL/DPEbEyIq7aweokSZJmlG5ejiQzvwZ8bdy093UMn9TN7UuSJE1X0+LGfEmSpCcaQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQVYAiTJEkqwBAmSZJUgCFMkiSpAEOYJElSAYYwSZKkAgxhkiRJBRjCJEmSCjCESZIkFWAIkyRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIUySJKkAQ5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIK6CldwFQYHh5m3bp1DAwMlC5lRujv72fBggX09vaWLkWSpBlrRoSwdevWMXfuXBYuXEhElC7ncS0z2bBhA+vWrePII48sXY4kSTPWjLgcOTAwwEEHHWQAmwIRwUEHHWSroiRJXTYjQhhgAJtCHktJkrpvxoQwSZKkxxND2BTYsGEDixcvZvHixcyfP5/DDjtsbHxoaGin1nHWWWdx66237vK2X/WqV/H85z9/u+mXXXYZxx9/PCeccAJLlizh4x//OFDd8/WRj3yEpz3taSxevJhnP/vZfP7zn9/l7UqSpD0zI27ML+2ggw5i5cqVAJx//vnsu+++nHvuudssk5lkJo3GxLn30ksv3eXtbty4kZtuuon+/n7Wrl3LU5/6VAD+5V/+hQsvvJBrrrmG+fPnMzAwwD/+4z8CcNFFF3HttdeyYsUK5s6dy0MPPcRXvvKVXd62JEnaMzMuhL3jHf/KypX3TOk6Fy+ezyc+cfIuv2716tWceuqpPOtZz+JHP/oR3/zmN/nABz7AD3/4Q7Zs2cLrX/963ve+9wHw/Oc/nwsvvJDjjz+egw8+mLe85S18/etfZ86cOXzlK1/hkEMO2W79V155Jaeffjr7778/l19+Oe9617sA+OAHP8jHPvYx5s+fD1Q/OfGmN71pbN5//ud/MnfuXAD2339/zjzzzN06LpIkafd5ObLLfvazn/HOd76TVatWcdhhh/GhD32IFStW8OMf/5hvfvObrFq1arvXPPTQQ7zwhS/kxz/+Mc95znO45JJLJlz38m1pTwUAABAjSURBVOXLOeOMMzjjjDNYvnz52PSbb76ZE088cbvlN27cyPDwMEccccTU7aAkSdotM64lbHdarLrpqKOOYunSpWPjy5cv5zOf+QwjIyP88pe/ZNWqVSxatGib18yePZtTTjkFgBNPPJHvf//72633l7/8JWvXruU5z3kOAO12m5/97Gcce+yxXdwbSZI0VWwJ67J99tlnbPi2227jk5/8JN/+9re56aabOPnkkyf8Pa6+vr6x4WazycjIyHbLfOELX+D+++9n4cKFLFy4kLVr1461hi1atIgbb7xxu9cceOCB9Pb2snbt2qnYNUmStAcMYXvRww8/zNy5c9lvv/24++67ufrqq3d7XcuXL+eaa65hzZo1rFmzhuuvv34shJ133nmce+653HvvvQAMDg7ymc98BoB3v/vd/MEf/AGbNm0aq+lzn/vcHu6ZJEnaVTPucuR0tmTJEhYtWsSxxx7LEUccwfOe97zdWs/tt9/O3Xffvc1lzmOOOYb+/n5uvPFGTj31VNavX89LXvISoPrx1Te/+c0AvO1tb+PRRx/lxBNPpK+vj97e3rEb+iVJ0t4TmVm6hl2ydOnSXLFixTbTbrnlFo477rhCFc1MHlNJkvZcRNyYmUsnmuflSEmSpAIMYZIkSQUYwiRJkgowhEmSJBVgCJMkSSrAECZJklSAIWwKbNiwgcWLF7N48WLmz5/PYYcdNjY+NDS00+u55JJLuOeeHT98fGhoiAMPPJD3vve920wfHh7mXe96F0cffTRLlizhuc997tgPwT788MO8+c1v5qijjuLEE0/kxS9+MTfccMPu7agkSZoy/ljrFDjooINYuXIlAOeffz777rsv55577i6v55JLLmHJkiXMnz9/wvlXX301ixYt4gtf+AJ/8Rd/MTb9vPPOY+PGjaxatYq+vj7uuece/u3f/g2As88+m+OOO47Vq1cTEdx+++38/Oc/3429lCRJU2nmhbBr3wH3rZzadR6yGF78id166Wc/+1kuuugihoaGeO5zn8uFF15Iu93mrLPOYuXKlWQm55xzDoceeigrV67k9a9/PbNnz+b666/f5hmSUD2q6I//+I/5+Mc/zvXXX8+yZcvYtGkTl112GWvWrBlbfv78+bzmNa/h1ltvZeXKlVxxxRVEBFA9UPyoo47as+MhSZL22MwLYdPIT3/6U770pS/xH//xH/T09HDOOedw+eWXc9RRR3H//ffzk5/8BIAHH3yQAw44gL/+67/mwgsvZPHixduta/PmzXznO98Zu2S5fPlyli1bxm233caRRx7Jvvvuu91rbr75Zp71rGfRaHjVWZKk6WbmhbDdbLHqhmuuuYYbbrhh7BmPW7Zs4fDDD+cVr3gFt956K29/+9v5jd/4DV7+8pc/5rquuuoqXvayl9Hf389rX/taTjzxRP7qr/6q27sgSZK6ZOaFsGkkMzn77LP58z//8+3m3XTTTXz961/noosu4otf/CIXX3zxpOtavnw51113HQsXLgRg/fr1fPe732Xp0qXceeedPPLII9u1hj396U9n5cqVtNttW8MkSZpm/GTuopNOOokrrriC+++/H6j+inLt2rWsX7+ezOS1r30tF1xwAT/84Q8BmDt3Lps2bdpuPQ8++CDXXXcd69atY82aNaxZs4ZPfepTLF++nLlz53LmmWfyjne8g+HhYQDuu+8+rrzySp72tKdxwgkncMEFFzD6oPY777yTr3/963vpCEiSpB0xhHXRCSecwPvf/35OOukknvGMZ/Dyl7+ce++9l7vuuosXvOAFLF68mLPOOosPfvCDAJx11lm86U1v2u6nLb74xS/yspe9jN7e3rFpp59+Ol/+8pcZHh7mQx/6EAcccADHHXccJ5xwAqeeeir7778/AJdeeil33XUXRx99NMcffzxnn302hxxyyN49EJIkaTsx2kLyeLF06dJcsWLFNtNuueUWjjvuuEIVzUweU0mS9lxE3JiZSyeaZ0uYJElSAYYwSZKkAmZMCHu8XVadzjyWkiR134wIYf39/WzYsMHwMAUykw0bNtDf31+6FEmSZrQZ8TthCxYsYN26daxfv750KTNCf38/CxYsKF2GJEkz2owIYb29vRx55JGly5AkSdppXb0cGREnR8StEbE6It49wfxZEfGFev4PImJhN+uRJEmaLroWwiKiCVwEnAIsAs6IiEXjFvs94IHMPBr4OPDhbtUjSZI0nXSzJWwZsDoz78jMIeBy4LRxy5wGfLYevhJ4aUREF2uSJEmaFrp5T9hhwF0d4+uAX9vRMpk5EhEPAQcB93cuFBHnAOfUo49ExK1dqfjx6WDGHS9NCY/r1POYdofHdep5TLvjiXpcj9jRjMfFjfmZeTFwcek6pqOIWLGjxyFo93lcp57HtDs8rlPPY9odHtftdfNy5C+AwzvGF9TTJlwmInqA/YENXaxJkiRpWuhmCLsBOCYijoyIPuANwFXjlrkKeGM9/Brg2+kvrkqSpCeArl2OrO/xeitwNdAELsnMmyPiAmBFZl4FfAb4XESsBjZSBTXtGi/TdofHdep5TLvD4zr1PKbd4XEdJ2x4kiRJ2vtmxLMjJUmSHm8MYZIkSQUYwh4HIqIZET+KiH+px4+sH/O0un7sU1893cdA7aSIOCAiroyIn0XELRHxnIg4MCK+GRG31f0n1ctGRHyqPq43RcSS0vVPVxHxzoi4OSJ+GhHLI6Lf83XXRMQlEXFfRPy0Y9oun5sR8cZ6+dsi4o0TbeuJZAfH9aP1e8BNEfGliDigY9559XG9NSJe0TF90sfxPZFMdEw75v1JRGREHFyPe65OwBD2+PBHwC0d4x8GPl4/7ukBqsc/gY+B2hWfBP41M48Fnkl1fN8NfCszjwG+VY9D9eitY+ruHOBv9365019EHAa8HViamcdT/UHOG/B83VWXASePm7ZL52ZEHAi8n+oHspcB7x8Nbk9gl7H9cf0mcHxmPgP4OXAeQP2IvTcAT69f8zf1l+GdeRzfE8llbH9MiYjDgZcDazsme65OwBA2zUXEAuA3gL+vxwN4CdVjnqB67NPp9bCPgdoJEbE/8AKqv84lM4cy80G2PX7jj+s/ZOU64ICIePJeLvvxogeYXf/u3xzgbjxfd0lmfo/qr8U77eq5+Qrgm5m5MTMfoAob231YPpFMdFwz8xuZOVKPXkf1e5ZQHdfLM3MwM+8EVlMFhJ15HN8Txg7OVai+VL0L6PzLP8/VCRjCpr9PUJ3M7Xr8IODBjjeOdVSPf4Jxj4ECRh8DpW0dCawHLq0v8/59ROwDHJqZd9fL3AMcWg9P9Aiuw9A2MvMXwP+i+vZ7N9X5dyOer1NhV89Nz9lddzbw9XrY47qbIuI04BeZ+eNxszymEzCETWMR8Srgvsy8sXQtM0wPsAT428x8FvAoWy/vAFD/aLC/37IL6ksIp1GF3KcA+/AE+ka7t3huTr2IeA8wAny+dC2PZxExB/gz4H2la3m8MIRNb88DTo2INVTN3i+hupfpgPpyD2z7OCgfA7Vz1gHrMvMH9fiVVKHs3tHLjHX/vnr+zjyCS3AScGdmrs/MYeD/UJ3Dnq97blfPTc/ZnRQRvwu8Cvjtjie2eFx3z1FUX8J+XH9uLQB+GBHz8ZhOyBA2jWXmeZm5IDMXUt0k+u3M/G3gWqrHPEH12Kev1MM+BmonZOY9wF0R8bR60kuBVWx7/MYf1zPrv+75deChjktD2mot8OsRMae+t2v0uHq+7rldPTevBl4eEU+qWyhfXk9Th4g4mep2j1Mzc3PHrKuAN9R/wXsk1c3k17Nzj+N7wsrMn2TmIZm5sP7cWgcsqd9zPVcnkpl2j4MOeBHwL/Xwr1C9IawG/hmYVU/vr8dX1/N/pXTd07UDFgMrgJuALwNPorof6VvAbcA1wIH1skH1F1G3Az+h+uu/4vswHTvgA8DPgJ8CnwNmeb7u8jFcTnVP3TDVh9jv7c65SXWP0+q6O6v0fpXudnBcV1Pdj7Sy7j7dsfx76uN6K3BKx/RXUv0l5e3Ae0rv13Q7puPmrwEOroc9VyfofGyRJElSAV6OlCRJKsAQJkmSVIAhTJIkqQBDmCRJUgGGMEmSpAIMYZKmRERkRPxVx/i5EXH+FK37soh4zWMvucfbeW1E3BIR13Z7W49Rx5qIOLhkDZK6zxAmaaoMAq+ebuGh49f6d8bvAW/OzBd3qx5JGmUIkzRVRoCLgXeOnzG+JSsiHqn7L4qI70bEVyLijoj4UET8dkRcHxE/iYijOlZzUkSsiIif189VJSKaEfHRiLghIm6KiN/vWO/3I+Iqql/tH1/PGfX6fxoRH66nvQ94PvCZiPjouOWfHBHfi4iV9Wv+Wz39b+uabo6ID3QsvyYi/me9/IqIWBIRV0fE7RHxlo4avxcRX42IWyPi0xGx3XtyRPxOfTxWRsT/rve5WR/Tn9b7sd0xlzT97co3REl6LBcBN0XER3bhNc8EjgM2AncAf5+ZyyLij4C3Ae+ol1sILKN6Pt21EXE0cCbV40+eHRGzgH+PiG/Uyy8Bjs/MOzs3FhFPAT4MnAg8AHwjIk7PzAsi4iXAuZm5YlyNvwVcnZl/GRFNYE49/T2ZubGe9q2IeEZm3lTPW5uZiyPi48BlVM/R7Kd6msCn62WWAYuA/wL+FXg11bNMR2s9Dng98LzMHI6IvwF+G7gZOCwzj6+XO2BnDrSk6cWWMElTJjMfBv4BePsuvOyGzLw7MwepHmkyGqJ+QhW8Rl2Rme3MvI0qrB1L9Zy5MyNiJfADqsf7HFMvf/34AFZ7NvCdrB40PgJ8HnjBY9UInFXf43ZCZm6qp78uIn4I/Ah4OlWgGjX6TMGfAD/IzE2ZuR4Y7AhN12fmHZnZonoEzPPHbfelVGHxhnofX0r1GKg7gF+JiL+un3/48GPUL2kasiVM0lT7BPBD4NKOaSPUX/rqS259HfMGO4bbHeNttn2PGv+MtaR6Ht3bMnObB/5GxIuAR3ev/O1l5vci4gXAbwCXRcTHgO8D5wLPzswHIuIyqpauUZ37MX4fR/dron3qFMBnM/O88TVFxDOBVwBvAV5H9fw9SY8jtoRJmlKZuRG4guom91FrqFp0AE4Fendj1a+NiEZ9n9ivUD1Y+Wrgf0REL0BE/GpE7PMY67keeGFEHFxfRjwD+O5kL4iII4B7M/PvgL+nutS5H1XQeygiDgVO2Y19WhYRR9bB9PXAv42b/y3gNRFxSF3HgRFxRP3HD43M/CLw3roeSY8ztoRJ6oa/At7aMf53wFci4sdU9z7tTivVWqoAtR/wlswciIi/p7pk+cOICGA9cPpkK8nMuyPi3cC1VC1NX83MrzzGtl8E/GlEDAOPAGdm5p0R8SPgZ8BdwL/vxj7dAFwIHF3X86Vxta6KiPdS3bfWAIaBPwS2AJd23Mi/XUuZpOkvMse3fkuSuq2+ZHpuZr6qdC2SyvBypCRJUgG2hEmSJBVgS5gkSVIBhjBJkqQCDGGSJEkFGMIkSZIKMIRJkiQV8H8BW8EIP2ssAlcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_acc = exponential_moving_average(score_trn_list[:,7], 0.1)\n",
    "tst_score_acc = exponential_moving_average(score_tst_list[:,7], 0.1)\n",
    "\n",
    "plt.title(\"ACC per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('ACC')\n",
    "plt.plot(train_size, trn_score_acc, label='Train ACC',color='navy')\n",
    "plt.plot(train_size, tst_score_acc, label='Test ACC',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### F1-score F1 = 2TP / (2TP + FP + FN):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trn_score_f1 = exponential_moving_average(score_trn_list[:,8], 0.1)\n",
    "tst_score_f1 = exponential_moving_average(score_tst_list[:,8], 0.1)\n",
    "\n",
    "plt.title(\"F1 per samples\")\n",
    "plt.xlabel('Number of samples')\n",
    "plt.ylabel('F1')\n",
    "plt.plot(train_size, trn_score_f1, label='Train F1',color='navy')\n",
    "plt.plot(train_size, tst_score_f1, label='Test F1',color=\"darkorange\")\n",
    "plt.legend(loc=\"best\")\n",
    "plt.ylim(0,1.05)\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}