cnn_grid_search.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import platform\n",
    "import random\n",
    "import shutil\n",
    "import sys\n",
    "\n",
    "import h5py\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import sklearn.metrics\n",
    "import tensorflow as tf\n",
    "from tensorflow.python.saved_model import tag_constants\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "import keras \n",
    "%matplotlib inline\n",
    "random.seed(0)\n",
    "\n",
    "from vis.visualization import visualize_saliency\n",
    "from vis.utils import utils\n",
    "from keras import activations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>index</th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>Unnamed: 0.1</th>\n",
       "      <th>On-Target Site</th>\n",
       "      <th>Off-Target Site</th>\n",
       "      <th>Score</th>\n",
       "      <th>Encoded Seqs</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>AAATGAGAAGAAGAGGCACAGGG</td>\n",
       "      <td>AAAAGAGAAGCTGAGGCACAGGG</td>\n",
       "      <td>1.0</td>\n",
       "      <td>KKKREKEKKEZREKEEPKPKEEE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>AAATGAGAAGAAGAGGCACAGGG</td>\n",
       "      <td>AAACGAGAAGAAGAGGCATAGGG</td>\n",
       "      <td>1.0</td>\n",
       "      <td>KKKMEKEKKEKKEKEEPKMKEEE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>AAATGAGAAGAAGAGGCACAGGG</td>\n",
       "      <td>AAACGAGAAGAAGCTGCACATGG</td>\n",
       "      <td>1.0</td>\n",
       "      <td>KKKMEKEKKEKKEZJEPKPKJEE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>AAATGAGAAGAAGAGGCACAGGG</td>\n",
       "      <td>AAATCATATGAAGAGGCACAAGG</td>\n",
       "      <td>1.0</td>\n",
       "      <td>KKKLHKJKREKKEKEEPKPKYEE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>AAATGAGAAGAAGAGGCACAGGG</td>\n",
       "      <td>AAATGAGAAAAATAGGCAGAGGG</td>\n",
       "      <td>1.0</td>\n",
       "      <td>KKKLEKEKKYKKJKEEPKHKEEE</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   index  Unnamed: 0  Unnamed: 0.1           On-Target Site  \\\n",
       "0      0           0             0  AAATGAGAAGAAGAGGCACAGGG   \n",
       "1      1           1             1  AAATGAGAAGAAGAGGCACAGGG   \n",
       "2      2           2             2  AAATGAGAAGAAGAGGCACAGGG   \n",
       "3      3           4             4  AAATGAGAAGAAGAGGCACAGGG   \n",
       "4      4           5             5  AAATGAGAAGAAGAGGCACAGGG   \n",
       "\n",
       "           Off-Target Site  Score             Encoded Seqs  \n",
       "0  AAAAGAGAAGCTGAGGCACAGGG    1.0  KKKREKEKKEZREKEEPKPKEEE  \n",
       "1  AAACGAGAAGAAGAGGCATAGGG    1.0  KKKMEKEKKEKKEKEEPKMKEEE  \n",
       "2  AAACGAGAAGAAGCTGCACATGG    1.0  KKKMEKEKKEKKEZJEPKPKJEE  \n",
       "3  AAATCATATGAAGAGGCACAAGG    1.0  KKKLHKJKREKKEKEEPKPKYEE  \n",
       "4  AAATGAGAAAAATAGGCAGAGGG    1.0  KKKLEKEKKYKKJKEEPKHKEEE  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# read in train, val, test dfs \n",
    "data_dir = './data/'\n",
    "# diff sheets, so need to read in \n",
    "file_name = data_dir + 'processed_binary_data' \n",
    "train_df = pd.read_csv(file_name+'_train.csv', sep = ',').reset_index()\n",
    "test_df = pd.read_csv(file_name+'_test.csv', sep = ',').reset_index()\n",
    "train_df.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Guide length:  23\n",
      "Off-target: 23\n"
     ]
    }
   ],
   "source": [
    "# check lengths -- confirm all are same when comparing across datasets\n",
    "seq_len = len(train_df['On-Target Site'][0])\n",
    "print('Guide length: ', seq_len) # NOTE: WT seq is the reverse complement of guide \n",
    "print('Off-target:', len(train_df['Off-Target Site'][0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/kcollins/anaconda/envs/deeptoehold/lib/python3.6/site-packages/ipykernel_launcher.py:19: FutureWarning: Method .as_matrix will be removed in a future version. Use .values instead.\n"
     ]
    }
   ],
   "source": [
    "from pysster.One_Hot_Encoder import One_Hot_Encoder\n",
    "\n",
    "from sklearn import preprocessing\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "new_alph_map = {'AA': 'K', 'AT': 'R', 'AC': 'Z', 'AG': 'Y', 'TT': 'L', 'TC': 'M', 'TG':'J', \n",
    "                'CC': 'P', 'CG':'H', 'GG': 'E'}\n",
    "\n",
    "new_alph = list(new_alph_map.values())\n",
    "alph_len = len(new_alph)\n",
    "\n",
    "# one-hot encode\n",
    "#modified code from Luis to get correct format for TPOT w/ my alphabet\n",
    "one = One_Hot_Encoder(''.join(new_alph))\n",
    "    \n",
    "# one-hot encode\n",
    "def _get_one_hot_encoding(seq):\n",
    "  \"\"\"Generate one-hot encoding for a single nucleotide sequence.\"\"\"\n",
    "  return pd.get_dummies(\n",
    "      list(seq) + new_alph)[:-len(new_alph)].as_matrix().transpose()\n",
    "\n",
    "# now convert the data into one_hot_encoding \n",
    "input_col_name = 'Encoded Seqs'\n",
    "train_data = np.stack(\n",
    "    [_get_one_hot_encoding(s) for s in train_df[input_col_name]]).astype(np.float32)\n",
    "#val_data = np.stack(\n",
    "    #[_get_one_hot_encoding(s) for s in val_df[input_col_name]]).astype(np.float32)\n",
    "test_data = np.stack(\n",
    "    [_get_one_hot_encoding(s) for s in test_df[input_col_name]]).astype(np.float32)\n",
    "train_data_f = np.reshape(train_data, (-1, alph_len * seq_len))\n",
    "#val_data_f = np.reshape(val_data, (-1, alph_len * seq_len))\n",
    "test_data_f = np.reshape(test_data, (-1, alph_len * seq_len))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import to_categorical\n",
    "\n",
    "output_bin_col_name = 'Score'; \n",
    "\n",
    "# NOTE: need one-hot encoding of the labels \n",
    "train_bins = to_categorical(np.array(train_df[output_bin_col_name].astype(np.float32)))\n",
    "#val_bins = to_categorical(np.array(val_df[output_bin_col_name].astype(np.float32)))\n",
    "test_bins = to_categorical(np.array(test_df[output_bin_col_name].astype(np.float32)))\n",
    "\n",
    "print('done')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.layers.core import Permute\n",
    "from keras import backend as K\n",
    "from keras.engine.topology import Layer\n",
    "import numpy as np\n",
    "import keras as keras\n",
    "from keras.callbacks import TensorBoard\n",
    "from keras import metrics as metrics\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten, Input, Embedding\n",
    "from keras.optimizers import SGD\n",
    "\n",
    "# need to reshape data for 1d conv\n",
    "num_features = seq_len\n",
    "num_channels = len(new_alph)\n",
    "train_data = train_data.reshape(train_data.shape[0], num_features, num_channels).astype('float32')\n",
    "test_data = test_data.reshape(test_data.shape[0], num_features, num_channels).astype('float32') \n",
    "\n",
    "from keras.layers import Conv1D, MaxPooling1D\n",
    "def conv1d(kernel_width, num_filters, hidden_layers,num_features = 23, num_channels = 10): \n",
    "    # already flattened b/c one_hot encoded\n",
    "    X_in = Input(shape=(num_features,num_channels),dtype='float32')\n",
    "    conv = Conv1D(filters=num_filters, kernel_size=kernel_width, padding='same')(X_in)\n",
    "    #pool = MaxPooling1D(pool_size=2)(conv)\n",
    "    H = Flatten()(conv)#(pool)\n",
    "    for h in hidden_layers: \n",
    "        H = Dense(h, activation='relu')(H)\n",
    "    out = Dense(2,activation=\"softmax\")(H)\n",
    "    model = Model(inputs=[X_in], outputs=out)\n",
    "    #model.summary()\n",
    "    #crossentropy is for probabilities\n",
    "    model.compile(loss='binary_crossentropy',optimizer='adam',metrics=['accuracy'])\n",
    "    print(model.summary())\n",
    "    return model\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_2 (Conv1D)            (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_8 (Dense)              (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 25,637\n",
      "Trainable params: 25,637\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7762728541561371\n",
      "AUC: 0.7248414637595851\n",
      "Accuracy: 0.9984590972878278\n",
      "F1: 0.7525553491925997\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_3 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_3 (Conv1D)            (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_3 (Flatten)          (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_10 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 25,637\n",
      "Trainable params: 25,637\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7639026834988489\n",
      "AUC: 0.7404655733966856\n",
      "Accuracy: 0.998349017665511\n",
      "F1: 0.7523338170043546\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_4 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_4 (Conv1D)            (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_4 (Flatten)          (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_16 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 25,637\n",
      "Trainable params: 25,637\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7246682885494325\n",
      "AUC: 0.6472627565145693\n",
      "Accuracy: 0.9985599633111671\n",
      "F1: 0.7078179848112822\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_5 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_5 (Conv1D)            (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_5 (Flatten)          (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_17 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_18 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_19 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_20 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 43,042\n",
      "Trainable params: 43,042\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7840206550772524\n",
      "AUC: 0.6282538750363438\n",
      "Accuracy: 0.9986608821668027\n",
      "F1: 0.7004845176387\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_6 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_6 (Conv1D)            (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_6 (Flatten)          (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_21 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_22 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_23 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_24 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 43,042\n",
      "Trainable params: 43,042\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.769984046560098\n",
      "AUC: 0.7457011754909263\n",
      "Accuracy: 0.9983673619136719\n",
      "F1: 0.7564217553242202\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_7 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_7 (Conv1D)            (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_7 (Flatten)          (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_25 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_26 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_27 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_28 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 43,042\n",
      "Trainable params: 43,042\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.8067477266941951\n",
      "AUC: 0.7786441571379175\n",
      "Accuracy: 0.998624168768631\n",
      "F1: 0.7924731497722627\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_8 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_8 (Conv1D)            (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_8 (Flatten)          (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_29 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_30 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_31 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_32 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 77,852\n",
      "Trainable params: 77,852\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7857312303318398\n",
      "AUC: 0.7353402323569023\n",
      "Accuracy: 0.9985508176873619\n",
      "F1: 0.7659093313138738\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_9 (InputLayer)         (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_9 (Conv1D)            (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_9 (Flatten)          (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_33 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_34 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_35 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_36 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 77,852\n",
      "Trainable params: 77,852\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.794218833193457\n",
      "AUC: 0.7380269426532359\n",
      "Accuracy: 0.9986975583805697\n",
      "F1: 0.7805381025867137\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_10 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_10 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_10 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_37 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_38 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_39 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_40 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 77,852\n",
      "Trainable params: 77,852\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8153785672428053\n",
      "AUC: 0.7235509813261504\n",
      "Accuracy: 0.998770924099977\n",
      "F1: 0.7792975651978863\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_11 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_11 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_11 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_41 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_42 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_43 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_44 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 25,737\n",
      "Trainable params: 25,737\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7988884142250401\n",
      "AUC: 0.7197758421864955\n",
      "Accuracy: 0.998780118686197\n",
      "F1: 0.7787643755528852\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_12 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_12 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_12 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_45 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_46 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_47 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_48 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 25,737\n",
      "Trainable params: 25,737\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7596401792075348\n",
      "AUC: 0.688316287420051\n",
      "Accuracy: 0.9985783207675234\n",
      "F1: 0.7404467677447494\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_13 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_13 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_13 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_49 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_50 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_51 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_52 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 25,737\n",
      "Trainable params: 25,737\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.7987519953463089\n",
      "AUC: 0.7445347012498398\n",
      "Accuracy: 0.9987067186425131\n",
      "F1: 0.7840798637878692\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_14 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_14 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_14 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_53 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_54 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_55 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_56 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 43,242\n",
      "Trainable params: 43,242\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7290854266053053\n",
      "AUC: 0.6386883266797116\n",
      "Accuracy: 0.9986241940069891\n",
      "F1: 0.7066868942207918\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_15 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_15 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_15 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_57 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_58 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_59 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_60 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 43,242\n",
      "Trainable params: 43,242\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.793170226356077\n",
      "AUC: 0.6779185889971778\n",
      "Accuracy: 0.9986883862564893\n",
      "F1: 0.7433992750367615\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_16 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_16 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_16 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_61 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_62 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_63 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_64 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 43,242\n",
      "Trainable params: 43,242\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.776278174140181\n",
      "AUC: 0.7129281892193156\n",
      "Accuracy: 0.9985416188947489\n",
      "F1: 0.7519713157954265\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_17 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_17 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_17 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_65 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_66 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_67 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_68 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 78,252\n",
      "Trainable params: 78,252\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7839563200239026\n",
      "AUC: 0.7066271140650836\n",
      "Accuracy: 0.9986150219670357\n",
      "F1: 0.7553166945599574\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_18 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_18 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_18 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_69 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_70 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_71 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_72 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 78,252\n",
      "Trainable params: 78,252\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7968285782886005\n",
      "AUC: 0.7823136372678079\n",
      "Accuracy: 0.998422394658155\n",
      "F1: 0.777955411082853\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_19 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_19 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_19 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_73 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_74 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_75 (Dense)             (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_76 (Dense)             (None, 2)                 22        \n",
      "=================================================================\n",
      "Total params: 78,252\n",
      "Trainable params: 78,252\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.7769003598692603\n",
      "AUC: 0.8019057095119846\n",
      "Accuracy: 0.9978628754872736\n",
      "F1: 0.7478447115729314\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_20 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_20 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_20 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_77 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_78 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_79 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 25,207\n",
      "Trainable params: 25,207\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7887909207968304\n",
      "AUC: 0.6517819259834529\n",
      "Accuracy: 0.9986792262467096\n",
      "F1: 0.7227463354574293\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_21 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_21 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_21 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_80 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_81 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_82 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 25,207\n",
      "Trainable params: 25,207\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.762900398025113\n",
      "AUC: 0.743101750885208\n",
      "Accuracy: 0.9983948782859134\n",
      "F1: 0.7572157826544426\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_22 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_22 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_22 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_83 (Dense)             (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_84 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_85 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 25,207\n",
      "Trainable params: 25,207\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7597022317269622\n",
      "AUC: 0.7129281892193156\n",
      "Accuracy: 0.9985416188947489\n",
      "F1: 0.7519713157954265\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_23 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_23 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_23 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_86 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_87 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_88 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 42,612\n",
      "Trainable params: 42,612\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7888328701640499\n",
      "AUC: 0.7144851112745839\n",
      "Accuracy: 0.9986517101268493\n",
      "F1: 0.7633280409500134\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_24 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_24 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_24 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_89 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_90 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_91 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 42,612\n",
      "Trainable params: 42,612\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7691963192093529\n",
      "AUC: 0.7118489321820316\n",
      "Accuracy: 0.9986058371397648\n",
      "F1: 0.7576127400938181\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_25 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_25 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_25 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_92 (Dense)             (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_93 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_94 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 42,612\n",
      "Trainable params: 42,612\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8044198390733762\n",
      "AUC: 0.7156792150357978\n",
      "Accuracy: 0.9987892685163953\n",
      "F1: 0.7767239327095863\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_26 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_26 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_26 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_95 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_96 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_97 (Dense)             (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 77,422\n",
      "Trainable params: 77,422\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.8034078676570279\n",
      "AUC: 0.6962523853075255\n",
      "Accuracy: 0.9987709466462436\n",
      "F1: 0.7637768429930272\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_27 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_27 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_27 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_98 (Dense)             (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_99 (Dense)             (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_100 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 77,422\n",
      "Trainable params: 77,422\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7902329477638887\n",
      "AUC: 0.7275419561337024\n",
      "Accuracy: 0.9986333535120063\n",
      "F1: 0.7690076767257482\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_28 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_28 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_28 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_101 (Dense)            (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_102 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_103 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 77,422\n",
      "Trainable params: 77,422\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7911882034363059\n",
      "AUC: 0.7392577610240514\n",
      "Accuracy: 0.9986608576014675\n",
      "F1: 0.7771037344333264\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_29 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_29 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_29 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_104 (Dense)            (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_105 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_106 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 25,307\n",
      "Trainable params: 25,307\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7954019903180982\n",
      "AUC: 0.6909754366000317\n",
      "Accuracy: 0.9986700542067561\n",
      "F1: 0.7505261757742144\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_30 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_30 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_30 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_107 (Dense)            (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_108 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_109 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 25,307\n",
      "Trainable params: 25,307\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7660658659352528\n",
      "AUC: 0.7118397439613305\n",
      "Accuracy: 0.9985874928916039\n",
      "F1: 0.7559754746544766\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_31 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_31 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_31 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_110 (Dense)            (None, 150)               17400     \n",
      "_________________________________________________________________\n",
      "dense_111 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_112 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 25,307\n",
      "Trainable params: 25,307\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7889797767896252\n",
      "AUC: 0.6946633363396546\n",
      "Accuracy: 0.9987892685163953\n",
      "F1: 0.7639826422372227\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_32 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_32 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_32 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_113 (Dense)            (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_114 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_115 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 42,812\n",
      "Trainable params: 42,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7502926073707412\n",
      "AUC: 0.7533570371214175\n",
      "Accuracy: 0.9980004952901574\n",
      "F1: 0.7349360707523347\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_33 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_33 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_33 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_116 (Dense)            (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_117 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_118 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 42,812\n",
      "Trainable params: 42,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.7755657481846518\n",
      "AUC: 0.72490577864518\n",
      "Accuracy: 0.9985874928916039\n",
      "F1: 0.7634500254696213\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_34 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_34 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_34 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_119 (Dense)            (None, 150)               34650     \n",
      "_________________________________________________________________\n",
      "dense_120 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_121 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 42,812\n",
      "Trainable params: 42,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7934556839063993\n",
      "AUC: 0.7025121320782546\n",
      "Accuracy: 0.9987250630589314\n",
      "F1: 0.76247947170846\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_35 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_35 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_35 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_122 (Dense)            (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_123 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_124 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 77,822\n",
      "Trainable params: 77,822\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7559236094469417\n",
      "AUC: 0.6778588669555943\n",
      "Accuracy: 0.9985691617672686\n",
      "F1: 0.7325185231130178\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_36 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_36 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_36 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_125 (Dense)            (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_126 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_127 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 77,822\n",
      "Trainable params: 77,822\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7782046586677025\n",
      "AUC: 0.719720711764795\n",
      "Accuracy: 0.9986700420083283\n",
      "F1: 0.7680376395768964\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_37 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_37 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_37 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_128 (Dense)            (None, 150)               69150     \n",
      "_________________________________________________________________\n",
      "dense_129 (Dense)            (None, 50)                7550      \n",
      "_________________________________________________________________\n",
      "dense_130 (Dense)            (None, 2)                 102       \n",
      "=================================================================\n",
      "Total params: 77,822\n",
      "Trainable params: 77,822\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8166295977970874\n",
      "AUC: 0.7917055755573985\n",
      "Accuracy: 0.998477413437285\n",
      "F1: 0.7857011490529107\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_38 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_38 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_38 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_131 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_132 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_133 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,847\n",
      "Trainable params: 14,847\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7800264411956503\n",
      "AUC: 0.6674657619254791\n",
      "Accuracy: 0.9986883982866629\n",
      "F1: 0.7358340531697883\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_39 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_39 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_39 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_134 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_135 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_136 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,847\n",
      "Trainable params: 14,847\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7801668811994789\n",
      "AUC: 0.6857260510350338\n",
      "Accuracy: 0.9986241813879259\n",
      "F1: 0.7428062690608236\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_40 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_40 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_40 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_137 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_138 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_139 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,847\n",
      "Trainable params: 14,847\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.7833285038134545\n",
      "AUC: 0.752273238340027\n",
      "Accuracy: 0.9984223801880303\n",
      "F1: 0.7633411992856329\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_41 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_41 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_41 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_140 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_141 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_142 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 26,502\n",
      "Trainable params: 26,502\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8093669933411427\n",
      "AUC: 0.7066822428827542\n",
      "Accuracy: 0.9987250864464765\n",
      "F1: 0.7656741028894393\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_42 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_42 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_42 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_143 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_144 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_145 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 26,502\n",
      "Trainable params: 26,502\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7628029901029145\n",
      "AUC: 0.6804674783890402\n",
      "Accuracy: 0.9985599765193623\n",
      "F1: 0.7335378115070421\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_43 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_43 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_43 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_146 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_147 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_148 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 26,502\n",
      "Trainable params: 26,502\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7648341483888927\n",
      "AUC: 0.7391107494928343\n",
      "Accuracy: 0.9983673469387755\n",
      "F1: 0.7523689395215418\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_44 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_44 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_44 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_149 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_150 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_151 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 49,812\n",
      "Trainable params: 49,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8239387232346547\n",
      "AUC: 0.7328970058990721\n",
      "Accuracy: 0.9988901831656378\n",
      "F1: 0.7973810276945132\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_45 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_45 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_45 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_152 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_153 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_154 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 49,812\n",
      "Trainable params: 49,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8051913015527007\n",
      "AUC: 0.8057911527055823\n",
      "Accuracy: 0.9983398455414305\n",
      "F1: 0.7815118808144722\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_46 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_46 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_46 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_155 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_156 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_157 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 49,812\n",
      "Trainable params: 49,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8025448416558181\n",
      "AUC: 0.7602828279408956\n",
      "Accuracy: 0.9986792020178858\n",
      "F1: 0.7891429660008142\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_47 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_47 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_47 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_158 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_159 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_160 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,947\n",
      "Trainable params: 14,947\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.7971394185480987\n",
      "AUC: 0.670088157040739\n",
      "Accuracy: 0.9987067423665698\n",
      "F1: 0.7395286817967843\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_48 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_48 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_48 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_161 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_162 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_163 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,947\n",
      "Trainable params: 14,947\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7755929702028186\n",
      "AUC: 0.7535270139701844\n",
      "Accuracy: 0.9983398455414305\n",
      "F1: 0.7582509130500696\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_49 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_49 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_49 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_164 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_165 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_166 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,947\n",
      "Trainable params: 14,947\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7728581110423153\n",
      "AUC: 0.6893266726793094\n",
      "Accuracy: 0.998624168768631\n",
      "F1: 0.7445535369863516\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_50 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_50 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_50 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_167 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_168 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_169 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 26,702\n",
      "Trainable params: 26,702\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8346677530263559\n",
      "AUC: 0.7956001911902032\n",
      "Accuracy: 0.9988626670457776\n",
      "F1: 0.8225723524995932\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_51 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_51 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_51 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_170 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_171 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_172 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 26,702\n",
      "Trainable params: 26,702\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7848711825275896\n",
      "AUC: 0.6831863498638725\n",
      "Accuracy: 0.9987709353732137\n",
      "F1: 0.7551667998074514\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_52 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_52 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_52 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_173 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_174 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_175 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 26,702\n",
      "Trainable params: 26,702\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8075207056568928\n",
      "AUC: 0.7130384478677283\n",
      "Accuracy: 0.998761751891768\n",
      "F1: 0.7724172884790648\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_53 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_53 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_53 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_176 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_177 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_178 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 50,212\n",
      "Trainable params: 50,212\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.797924340594026\n",
      "AUC: 0.7144988934790016\n",
      "Accuracy: 0.9986792262467096\n",
      "F1: 0.7659031057394288\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_54 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_54 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_54 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_179 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_180 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_181 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 50,212\n",
      "Trainable params: 50,212\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.7916727705844138\n",
      "AUC: 0.7457654930358338\n",
      "Accuracy: 0.998495771650799\n",
      "F1: 0.7666687894099469\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_55 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_55 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_55 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_182 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_183 (Dense)            (None, 30)                3030      \n",
      "_________________________________________________________________\n",
      "dense_184 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 50,212\n",
      "Trainable params: 50,212\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8136283163375027\n",
      "AUC: 0.7524570027540483\n",
      "Accuracy: 0.9987892685163953\n",
      "F1: 0.7959931630003187\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_56 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_56 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_56 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_185 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_186 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 11,957\n",
      "Trainable params: 11,957\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7979267026099166\n",
      "AUC: 0.722292573734553\n",
      "Accuracy: 0.9985875058471755\n",
      "F1: 0.7619920299975108\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_57 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_57 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_57 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_187 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_188 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 11,957\n",
      "Trainable params: 11,957\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8142581887079423\n",
      "AUC: 0.7406493378107069\n",
      "Accuracy: 0.9987159026287308\n",
      "F1: 0.7836291152263375\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_58 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_58 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_58 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_189 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_190 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 11,957\n",
      "Trainable params: 11,957\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8194724627913781\n",
      "AUC: 0.6920133809508842\n",
      "Accuracy: 0.9987434074753497\n",
      "F1: 0.7576359736516779\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_59 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_59 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_59 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_191 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_192 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 23,612\n",
      "Trainable params: 23,612\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7976588774843772\n",
      "AUC: 0.7066454903376405\n",
      "Accuracy: 0.9986517101268493\n",
      "F1: 0.7586788488385146\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_60 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_60 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_60 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_193 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_194 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 23,612\n",
      "Trainable params: 23,612\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7722950673758366\n",
      "AUC: 0.750909212923064\n",
      "Accuracy: 0.99833067341735\n",
      "F1: 0.7562664433012054\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_61 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_61 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_61 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_195 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_196 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 23,612\n",
      "Trainable params: 23,612\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8213366924156984\n",
      "AUC: 0.7051620874670251\n",
      "Accuracy: 0.998770924099977\n",
      "F1: 0.7686578390635689\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_62 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_62 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_62 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_197 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_198 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 46,922\n",
      "Trainable params: 46,922\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.8272518420676166\n",
      "AUC: 0.7590152934844666\n",
      "Accuracy: 0.9988626670457776\n",
      "F1: 0.8071686624151202\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_63 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_63 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_63 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_199 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_200 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 46,922\n",
      "Trainable params: 46,922\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7814391418859885\n",
      "AUC: 0.8186734229754106\n",
      "Accuracy: 0.9979729605782107\n",
      "F1: 0.7618577485579738\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_64 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_64 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_64 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_201 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_202 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 46,922\n",
      "Trainable params: 46,922\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7987747443772564\n",
      "AUC: 0.736644558518085\n",
      "Accuracy: 0.9986883742260949\n",
      "F1: 0.7783093780021423\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_65 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_65 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_65 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_203 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_204 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 12,057\n",
      "Trainable params: 12,057\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8215160628479463\n",
      "AUC: 0.7719894470165882\n",
      "Accuracy: 0.9986792262467096\n",
      "F1: 0.7951238181300377\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_66 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_66 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_66 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_205 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_206 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 12,057\n",
      "Trainable params: 12,057\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8082864722134869\n",
      "AUC: 0.7484062646347068\n",
      "Accuracy: 0.9985508043952819\n",
      "F1: 0.7726256277108937\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_67 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_67 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_67 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_207 (Dense)            (None, 100)               11600     \n",
      "_________________________________________________________________\n",
      "dense_208 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 12,057\n",
      "Trainable params: 12,057\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8249097543823865\n",
      "AUC: 0.7367042819526419\n",
      "Accuracy: 0.9988076129328136\n",
      "F1: 0.7900240594731212\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_68 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_68 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_68 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_209 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_210 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 23,812\n",
      "Trainable params: 23,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8339533266688834\n",
      "AUC: 0.7799347315207338\n",
      "Accuracy: 0.9988901831656378\n",
      "F1: 0.8191251039526302\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_69 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_69 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_69 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_211 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_212 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 23,812\n",
      "Trainable params: 23,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7921430116625591\n",
      "AUC: 0.7145356448844099\n",
      "Accuracy: 0.9987525911250528\n",
      "F1: 0.7730210514713987\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_70 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_70 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_70 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_213 (Dense)            (None, 100)               23100     \n",
      "_________________________________________________________________\n",
      "dense_214 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 23,812\n",
      "Trainable params: 23,812\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.8025572841890762\n",
      "AUC: 0.7968906335761764\n",
      "Accuracy: 0.9983398303141481\n",
      "F1: 0.7772254590760125\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_71 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_71 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_71 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_215 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_216 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 47,322\n",
      "Trainable params: 47,322\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8028122881674101\n",
      "AUC: 0.7485486833641238\n",
      "Accuracy: 0.9988351509259175\n",
      "F1: 0.7993929063466998\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_72 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_72 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_72 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_217 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_218 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 47,322\n",
      "Trainable params: 47,322\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7809903314046441\n",
      "AUC: 0.7535545786322876\n",
      "Accuracy: 0.9983948782859134\n",
      "F1: 0.7624706680770864\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_73 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_73 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_73 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_219 (Dense)            (None, 100)               46100     \n",
      "_________________________________________________________________\n",
      "dense_220 (Dense)            (None, 2)                 202       \n",
      "=================================================================\n",
      "Total params: 47,322\n",
      "Trainable params: 47,322\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8044103361969452\n",
      "AUC: 0.6920317573922863\n",
      "Accuracy: 0.9987800963081862\n",
      "F1: 0.7613433788730982\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_74 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_74 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_74 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_221 (Dense)            (None, 30)                3480      \n",
      "_________________________________________________________________\n",
      "dense_222 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 3,697\n",
      "Trainable params: 3,697\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.822353247513838\n",
      "AUC: 0.6884265465749996\n",
      "Accuracy: 0.9987984627661038\n",
      "F1: 0.7615174182000185\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_75 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_75 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_75 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_223 (Dense)            (None, 30)                3480      \n",
      "_________________________________________________________________\n",
      "dense_224 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 3,697\n",
      "Trainable params: 3,697\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8190240387290679\n",
      "AUC: 0.7459125045670508\n",
      "Accuracy: 0.9987892796213748\n",
      "F1: 0.79344687505741\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_76 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_76 (Conv1D)           (None, 23, 5)             155       \n",
      "_________________________________________________________________\n",
      "flatten_76 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_225 (Dense)            (None, 30)                3480      \n",
      "_________________________________________________________________\n",
      "dense_226 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 3,697\n",
      "Trainable params: 3,697\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8242171317715772\n",
      "AUC: 0.7524432204229967\n",
      "Accuracy: 0.998761751891768\n",
      "F1: 0.793267954690513\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_77 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_77 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_77 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_227 (Dense)            (None, 30)                6930      \n",
      "_________________________________________________________________\n",
      "dense_228 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 7,302\n",
      "Trainable params: 7,302\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8530917439307213\n",
      "AUC: 0.779962295929569\n",
      "Accuracy: 0.9989452154053583\n",
      "F1: 0.8249638689110694\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_78 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_78 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_78 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_229 (Dense)            (None, 30)                6930      \n",
      "_________________________________________________________________\n",
      "dense_230 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 7,302\n",
      "Trainable params: 7,302\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.8100564718249725\n",
      "AUC: 0.805814123257335\n",
      "Accuracy: 0.998385706161833\n",
      "F1: 0.7849615199361852\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_79 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_79 (Conv1D)           (None, 23, 10)            310       \n",
      "_________________________________________________________________\n",
      "flatten_79 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_231 (Dense)            (None, 30)                6930      \n",
      "_________________________________________________________________\n",
      "dense_232 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 7,302\n",
      "Trainable params: 7,302\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8229270145064216\n",
      "AUC: 0.7078166369661463\n",
      "Accuracy: 0.9988259573492319\n",
      "F1: 0.7759298800793263\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_80 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_80 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_80 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_233 (Dense)            (None, 30)                13830     \n",
      "_________________________________________________________________\n",
      "dense_234 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,512\n",
      "Trainable params: 14,512\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7986424430052448\n",
      "AUC: 0.7223017618708315\n",
      "Accuracy: 0.9986058499270822\n",
      "F1: 0.7636261023153086\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_81 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_81 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_81 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_235 (Dense)            (None, 30)                13830     \n",
      "_________________________________________________________________\n",
      "dense_236 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,512\n",
      "Trainable params: 14,512\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.7882293742811759\n",
      "AUC: 0.7720261974933476\n",
      "Accuracy: 0.9987525911250528\n",
      "F1: 0.8020132364195096\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_82 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_82 (Conv1D)           (None, 23, 20)            620       \n",
      "_________________________________________________________________\n",
      "flatten_82 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_237 (Dense)            (None, 30)                13830     \n",
      "_________________________________________________________________\n",
      "dense_238 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,512\n",
      "Trainable params: 14,512\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8245350871208565\n",
      "AUC: 0.731431935837204\n",
      "Accuracy: 0.998770924099977\n",
      "F1: 0.7835632613030634\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_83 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_83 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_83 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_239 (Dense)            (None, 30)                3480      \n",
      "_________________________________________________________________\n",
      "dense_240 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 3,797\n",
      "Trainable params: 3,797\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8119934875292623\n",
      "AUC: 0.821433646550966\n",
      "Accuracy: 0.9982664844488063\n",
      "F1: 0.782324375518048\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_84 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_84 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_84 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_241 (Dense)            (None, 30)                3480      \n",
      "_________________________________________________________________\n",
      "dense_242 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 3,797\n",
      "Trainable params: 3,797\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8117483729861718\n",
      "AUC: 0.7744694223471477\n",
      "Accuracy: 0.9984132225340745\n",
      "F1: 0.7737540436603463\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_85 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_85 (Conv1D)           (None, 23, 5)             255       \n",
      "_________________________________________________________________\n",
      "flatten_85 (Flatten)         (None, 115)               0         \n",
      "_________________________________________________________________\n",
      "dense_243 (Dense)            (None, 30)                3480      \n",
      "_________________________________________________________________\n",
      "dense_244 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 3,797\n",
      "Trainable params: 3,797\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8316894083909043\n",
      "AUC: 0.7393312667896599\n",
      "Accuracy: 0.9988076129328136\n",
      "F1: 0.7913681427526048\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_86 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_86 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_86 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_245 (Dense)            (None, 30)                6930      \n",
      "_________________________________________________________________\n",
      "dense_246 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 7,502\n",
      "Trainable params: 7,502\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average precision-recall score: 0.8039300909135196\n",
      "AUC: 0.6544089151668517\n",
      "Accuracy: 0.9987067423665698\n",
      "F1: 0.7274755289080687\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_87 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_87 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_87 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_247 (Dense)            (None, 30)                6930      \n",
      "_________________________________________________________________\n",
      "dense_248 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 7,502\n",
      "Trainable params: 7,502\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8078175840142116\n",
      "AUC: 0.7380223485428854\n",
      "Accuracy: 0.9986883862564893\n",
      "F1: 0.7796716071043095\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_88 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_88 (Conv1D)           (None, 23, 10)            510       \n",
      "_________________________________________________________________\n",
      "flatten_88 (Flatten)         (None, 230)               0         \n",
      "_________________________________________________________________\n",
      "dense_249 (Dense)            (None, 30)                6930      \n",
      "_________________________________________________________________\n",
      "dense_250 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 7,502\n",
      "Trainable params: 7,502\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8082252908100369\n",
      "AUC: 0.718283229321063\n",
      "Accuracy: 0.9987434074753497\n",
      "F1: 0.7736128074637018\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_89 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_89 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_89 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_251 (Dense)            (None, 30)                13830     \n",
      "_________________________________________________________________\n",
      "dense_252 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,912\n",
      "Trainable params: 14,912\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.826328743660842\n",
      "AUC: 0.6806007078424735\n",
      "Accuracy: 0.998825978885964\n",
      "F1: 0.759104632483186\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_90 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_90 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_90 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_253 (Dense)            (None, 30)                13830     \n",
      "_________________________________________________________________\n",
      "dense_254 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,912\n",
      "Trainable params: 14,912\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.794391280152647\n",
      "AUC: 0.743161474319765\n",
      "Accuracy: 0.9985141158989599\n",
      "F1: 0.7668693144883847\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_91 (InputLayer)        (None, 23, 10)            0         \n",
      "_________________________________________________________________\n",
      "conv1d_91 (Conv1D)           (None, 23, 20)            1020      \n",
      "_________________________________________________________________\n",
      "flatten_91 (Flatten)         (None, 460)               0         \n",
      "_________________________________________________________________\n",
      "dense_255 (Dense)            (None, 30)                13830     \n",
      "_________________________________________________________________\n",
      "dense_256 (Dense)            (None, 2)                 62        \n",
      "=================================================================\n",
      "Total params: 14,912\n",
      "Trainable params: 14,912\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Average precision-recall score: 0.8089426793231191\n",
      "AUC: 0.7209285905994831\n",
      "Accuracy: 0.9987800963081862\n",
      "F1: 0.7787643699429052\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import precision_recall_curve, average_precision_score\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.utils.fixes import signature\n",
    "\n",
    "# fix random seed for reproducibility\n",
    "seed = 7\n",
    "np.random.seed(seed)\n",
    "# define k-fold cross validation test harness\n",
    "num_folds = 3\n",
    "num_epochs = 20\n",
    "kfold = StratifiedKFold(n_splits=num_folds, shuffle=True, random_state=seed)\n",
    "all_scores = {}\n",
    "X = train_data\n",
    "y = train_bins\n",
    "\n",
    "kernel_widths = [3,5]\n",
    "num_filters = [5,10,20]\n",
    "hidden_layer_choices = [(150,50,10), (150,50),(100,30), (100,),(30,)]\n",
    "\n",
    "for hidden_layers in hidden_layer_choices: \n",
    "    for kernel_width in kernel_widths: \n",
    "        for num_filt in num_filters:\n",
    "            param_combo = (hidden_layers,kernel_width,num_filt)\n",
    "            cvscores = []\n",
    "            for train, test in kfold.split(X, y.argmax(1)):\n",
    "              # create model\n",
    "                model = conv1d(kernel_width = kernel_width, num_filters = num_filt, hidden_layers=hidden_layers)\n",
    "                #early_stopping = keras.callbacks.EarlyStopping(monitor='val_loss', min_delta=0.005, patience=10, verbose=0, mode='auto', baseline=None)\n",
    "                model.fit(X[train], y[train],epochs=num_epochs, batch_size=128,verbose=0)#,callbacks=[early_stopping])\n",
    "                # evaluate the model\n",
    "\n",
    "                y_pred = model.predict(X[test])\n",
    "                y_true = y[test]\n",
    "\n",
    "                y_true_labels = np.argmax(y[test], axis=1)\n",
    "                y_pred_labels = np.argmax(y_pred, axis=1) \n",
    "\n",
    "                average_precision = average_precision_score(y_true, y_pred)\n",
    "                auc = roc_auc_score(y_true_labels, y_pred_labels)\n",
    "                acc = sklearn.metrics.accuracy_score(y_true_labels, y_pred_labels)\n",
    "                f1_val = sklearn.metrics.f1_score(y_true_labels,y_pred_labels, average='macro')\n",
    "                print('Average precision-recall score:', average_precision)\n",
    "                print('AUC:', auc)\n",
    "                print('Accuracy:', acc)\n",
    "                print('F1:', f1_val)\n",
    "                cvscores.append([average_precision, auc, acc, f1_val])\n",
    "            all_scores[param_combo] = cvscores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "d = pd.DataFrame.from_dict(all_scores)\n",
    "d.to_csv('cnn_grid_search_all.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "cv_means = {k:sum(np.array(v)/num_folds) for k,v in all_scores.items()}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<BarContainer object of 30 artists>"
      ]
     },
     "execution_count": 12,
     "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": [
    "cv_means_df = pd.DataFrame.from_dict(cv_means)\n",
    "param_combos = list(cv_means.keys())\n",
    "score_vals = [cv_means[k][0] for k in param_combos]\n",
    "plt.bar(range(len(param_combos)),score_vals)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8286917434207051"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max(score_vals)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1998eaf28>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAACxUAAAJCCAYAAAAMIgu0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3W+MZfddHvDnjO21Bxuvk93B8bWzMVwbicCbSoFSSCpFUVo3UuyWViiBqCAhgoZCJcQtDYmJUmIbGt0StSq5Iq2qRECJ/CLItuoS+sdScUNUm1JMsnbInqBdb05s1uN4x95c7+x6T1/szv6zE493zu7vzJzPR4rmzp1zv3nmvrh7fec531O1bRsAAAAAAAAAAAAAYLgWSgcAAAAAAAAAAAAAAMpSKgYAAAAAAAAAAACAgVMqBgAAAAAAAAAAAICBUyoGAAAAAAAAAAAAgIFTKgYAAAAAAAAAAACAgVMqBgAAAAAAAAAAAICBUyoGAAAAAAAAAAAAgIFTKgYAAAAAAAAAAACAgVMqBgAAAAAAAAAAAICBu7x0gPPt3r27vfnmm0vHAAAAAAAAAAAAAIAt7c/+7M+eadt2aSPH9q5UfPPNN+fRRx8tHQMAAAAAAAAAAAAAtrSqqvZv9NiFixkEAAAAAAAAAAAAAOg/pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABg4pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABg4pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABg4pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABg4pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABg4pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABg4pWIAAAAAAAAAAAAAGDilYgAAAAAAAAAAAAAYOKViAAAAAAAAAAAAABi4y0sHAAAAAAAAAAAA6IvZbJa6rjuZ1TRNkmQ0GnUybzweZ3l5uZNZAHA+pWIAAAAAAAAAAICLYD6fl44AABumVAwAAAAAAAAAAHBKl5uAJ5NJkmQ6nXY2EwAuloXSAQAAAAAAAAAAAACAspSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOCUigEAAAAAAAAAAABg4JSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOA2VCququq2qqq+XFXVvqqqPvAKP99TVdVDVVX9eVVVj1VV9a5X+PkLVVVNugoOAAAAAAAAAAAAAHTjVUvFVVVdluS3k/yDJG9O8t6qqt583mF3Jrm3bdu/leQ9ST5x3s9/K8l/3XxcAAAAAAAAAAAAAKBrG9lU/ENJ9rVt+9W2bdeSfCbJHecd0ya59tTtnUma9R9UVfUPk/x1ki9tPi4AAAAAAAAAAAAA0LWNlIpvTPLkWd8fPHXf2T6S5H1VVR1M8mCSX0ySqqquSfIvk/yrb/d/UFXV+6uqerSqqkcPHTq0wegAAAAAAAAAAAAAQBc2UireiPcm+VTbtjcleVeS362qaiEny8Yfb9v2hW/34LZtP9m27Vvatn3L0tJSR5EAAAAAAAAAAAAAgI24fAPHfC3JG8/6/qZT953tZ5LcliRt2/5pVVVXJdmd5G8n+SdVVX0syXVJTlRV9WLbtv9+08kBAAAAAAAAAAAAgE5spFT8SJJbq6r67pwsE78nyU+cd8yBJO9I8qmqqr4vyVVJDrVt+7b1A6qq+kiSFxSKAQAAAAAAAAAAAKBfFl7tgLZtjyf5hSSfS/J4knvbtv1SVVW/XlXV7acO++UkP1tV1V8k+YMkP922bXuxQgMAAAAAAAAAAAAA3dnIpuK0bftgkgfPu+/DZ93em+RHX2XGRy4gHwAAAAAAAAAAAABwkb3qpmIAAAAAAAAAAAAAYHvb0KZiAAAAAAAAAACAPpvNZqnrunSMc6znmUwmhZOcazweZ3l5uXQMAHpGqRgAAAAAAAAAANjy6rrOV/buyxuv3VM6ymlXHN+RJHnx4FrhJGc8uXqgdAQAekqpGAAAAAAAAAAA2BbeeO2eTH74g6Vj9Nr0C/eUjgBATy2UDgAAAAAAAAAAAAAAlKVUDAAAAAAAAAAAAAADp1QMAAAAAAAAAAAAAAOnVAwAAAAAAAAAAAAAA6dUDAAAAAAAAAAAAAADp1QMAAAAAAAAAAAAAAOnVAwAAAAAAAAAAAAAA6dUDAAAAAAAAAAAAAADp1QMAAAAAAAAAAAAAAOnVAwAAAAAAAAAAAAAA3d56QAAAAAAAEC/zGaz1HXdyaymaZIko9Gok3nj8TjLy8udzAIAAAAAzlAqBgAAAAAALpr5fF46AgAAAACwAUrFAAAAAADAObrcBDyZTJIk0+m0s5kAAAAAQPeUigEAAAAAAACALWk2m6Wu605mNU2TJBmNRp3MG4/HnZ6sBQAAF5tSMQAAAAAAAAAwePP5vHQEAAAoSqkYAAAAAAAAANiSutwEPJlMkiTT6bSzmQAAsJUslA4AAAAAAAAAAAAAAJSlVAwAAAAAAAAAAAAAA6dUDAAAAAAAAAAAAAADp1QMAAAAAAAAAAAAAAOnVAwAAAAAAAAAAAAAA3d56QAAAAAAAAAAAADApTObzVLXdSezmqZJkoxGo07mjcfjLC8vdzILeG2UigEAAAAAAAAAAIALMp/PS0cAOqJUDAAAAAAAAAAAAAPS5SbgyWSSJJlOp53NBMpYKB0AAAAAAAAAAAAAAChLqRgAAAAAAAAAAAAABk6pGAAAAAAAAAAAAAAGTqkYAAAAAAAAAAAAAAZOqRgAAAAAAAAAAAAABk6pGAAAAAAAAAAAAAAGTqkYAAAAAAAAAAAAAAZOqRgAAAAAAAAAAAAABk6pGAAAAAAAAAAAAAAGTqkYAAAAAAAAAAAAAAZOqRgAAAAAAAAAAAAABk6pGAAAAAAAAAAAAAAGTqkYAAAAAAAAAAAAAAZOqRgAAAAAAAAAAAAABu7y0gEAAAAAAAAA6N5sNktd153MapomSTIajTqZNx6Ps7y83MksAAAAuqFUDAAAAAAAAMC3NZ/PS0cAAADgIlMqBgAAAAAAANiGutwEPJlMkiTT6bSzmQAAAPTLQukAAAAAAAAAAAAAAEBZNhUDAAAAAAAAAAAAsK3MZrPUdd3JrKZpkiSj0aiTeePxuNOry3RFqRgAAAAAAAAAAAAAvoX5fF46wiWhVAwAAAAAAAAAAADAttLlJuDJZJIkmU6nnc3so4XSAQAAAAAAAAAAAACAspSKAQAAAAAAAAAAAGDgLi8dAAAAAAAAAAC2g9lslrquO5nVNE2SZDQadTJvPB53evlnAABg+1EqBgAAAAAAAICemc/npSMAAAADo1QMAAAAAAAAAB3ochPwZDJJkkyn085mAkBf2fYP0A9KxQAAAAAAAAAA0FPKlvDa2PYPcOGUigEAAAAAXgN/zAUAAGCrUrakr2z7B+gHpWIAAAAAgEL8MRcAAIBXo2wJAFwqSsUAAAAAAK+BP+YCAAAAALAdKRUDAAAAAAAAMGiz2Sx1XZeOcY71POsnovXFeDzu9EQ72M68tmyc1xYA6AelYgAAAAAAAAAGra7r7H1iX3bu3lM6ymkvZUeS5GvPrBVOcsbhZw6UjgBbSl3X+avH92V0XX9eWy47cfK15YWv9+e1pXmuu9eWpmlyZPVIpl+4p7OZ29GTq/tzdXN16RgA9JBSMQAAAAAAAACDt3P3nrz1jg+VjtFrD993d+kIsOWMrtuTn3v7naVj9NrvPHRX6QgAwClKxQAAAAAAAAAA0LGmafLC4SNKs6+ieW5/rmm72Zo7Go3y4om1TH74g53M266mX7gnV412lI4BQA8tlA4AAAAAAAAAAAAAAJRlUzEAAAAAAAAAAHRsNBrlhWotP/f2O0tH6bXfeeiuXHODrbmwUbPZLHVdl45xjvU8k8mkcJJzjcfjLC8vl44BW4pSMQAAAAAAAACD1jRNDq8eycP33V06Sq8dfmZ/qrWrS8cAgEGr6zr79v5V9lx7Q+kop+04flmSZO3g84WTnHFg9eulI8CWpFQMAAAAAAAAXHRdblRrmibJyQ2QXbDBDACArWTPtTfkQz/8/tIxeu3uL3yydATYkpSKAQAAAABgm3AJ1I1TIIStbT6fl47ANjMajdLuWMtb7/hQ6Si99vB9d2e0e0fpGAAAwEWiVAwAAAAAANtEXdd57IknUu3aVTrKaW3bJkn+8tChwknOaFdWSkeAQeqyyL9+osJ0Ou1sJgAAAAzdhkrFVVXdluTfJrksyX9s2/Y3z/v5niSfTnLdqWM+0Lbtg1VVvTPJbybZkWQtyb9o2/Z/dpgfAAAAAAA4S7VrV3a8+47SMXpt7YH7SkcAAGAgmucO5Hceuqt0jNOeeeHpJMnua64vnOSM5rkD+d4bbikdAwDIBkrFVVVdluS3k7wzycEkj1RVdX/btnvPOuzOJPe2bTurqurNSR5McnOSZ5K8u23bpqqqH0jyuSQ3dvw7AAAAAAAAAABAr4zH49IRXubpei1Jcs0NOwonOeN7b7ill88VAAzRRjYV/1CSfW3bfjVJqqr6TJI7kpxdKm6TXHvq9s4kTZK0bfvnZx3zpSSLVVVd2bbt0c0GBwAAAAAAAACAvlpeXi4d4WUmk0mSZDqdFk4CAPTRwgaOuTHJk2d9fzAv3zb8kSTvq6rqYE5uKf7FV5jzj5P831cqFFdV9f6qqh6tqurRQ4cObSg4AAAAAAAAAAAAANCNjWwq3oj3JvlU27b/pqqqv5Pkd6uq+oG2bU8kSVVV35/kXyf5e6/04LZtP5nkk0nylre8pe0oEwAAAAAAALyi2WyWuq47mdU0TZJkNBp1Mm88HvdysyEAAACwvW2kVPy1JG886/ubTt13tp9JcluStG37p1VVXZVkd5K/qarqpiR/mOSftm3bzSczAAAAAAAA0BPz+bx0BAAAAIBN20ip+JEkt1ZV9d05WSZ+T5KfOO+YA0nekeRTVVV9X5Krkhyqquq6JP8lyQfatv3f3cUGAAAAAACAC9flJuDJZJIkmU6nnc0EAFjX5RUW1uesv3/ZLFdYAIDtZeHVDmjb9niSX0jyuSSPJ7m3bdsvVVX161VV3X7qsF9O8rNVVf1Fkj9I8tNt27anHndLkg9XVfX/Tv3vuy7KbwIAAAAAAAAAAHxLi4uLWVxcLB0DAOipjWwqTtu2DyZ58Lz7PnzW7b1JfvQVHndXkrs2mREAAAAAAAAAAAbJJmAA4FLZUKkYAAAAAAAAAKArs9ksdV2XjnGO9TyTyaRwknONx2OlUgAALgmlYgAAAAAAAADgkqrrOk88sS+7dr2pdJTT2nZHkuTQoWOFk5yxsrK/dAQAAAZEqRgAAAAAAAAAuOR27XpT7nj3r5WO0Wv3PfDR0hEAAC45V7XYuK6vaqFUDAAAAAAAAAAAAEAv1HWdfXufyJ6dS6WjnLbjpZNf1762UjbIWQ4cPtT5TKViAAAAAAAAAAAAAHpjz86l3PnWHysdo9fueviznc9UKgYAAAAAAGDLcAnUjev6EqgAAADA9qZUDAAAAGxbXRZOmqZJkoxGo07mKXgAAFyYuq7z2BNfzsKuN5SOctqJdiFJ8sVDhwsnOePEylOlIwAAAABbjFIxAAAAwAbM5/PSEQAAOGVh1xuyePtPlY7Ra/P7P106AgBAEU+uHsj0C/eUjnHa3xx5OknyXVdfXzjJGU+uHsituaV0DAB6SKkYAAAA2La63AS8finr6XTa2UwAAAAAoDvj8bh0hJc5Vq8lSa66aUfhJGfcmlt6+VwBUJ5SMQAAAAAAAAAAsOV1uWSgK5YVALCVLJQOAAAAAAAAAAAAAACUZVMxbHOz2Sx1XXcyq2maJMloNOpk3ng87uVZggAAAAAAAAAAADA0SsXAhs3n89IRAAAAAAAAAAAAgItAqRi2uS43AU8mkyTJdDrtbCYAAAAAAAD0weFnDuTh++4uHeO0I4efTpJcvfP6wknOOPzMgdy4+5bSMQAAgItEqRgAAAAAGITZbJa6rkvHOMd6nvWTuftiPB53erI6AEDfjcfj0hFepj68liS5cfeOwknOuHH3Lb18rgAAgG4oFQMAAAAAg1DXdR574vFUu64rHeW0tj2RJPnLQ18vnOSMduW50hEAvq2maXJi9fnM7/906Si9dmLlqTTHjpSOAVtGH0+ochVRAADgUlMqBgAAAAAGo9p1XS6//e2lY/Ta8fsfKh0BAAAAAIAClIoBAAAAAADYMkajUZ694nAWb/+p0lF6bX7/pzNa2lk6BgAAALCFKBUDAAAAAAAUNpvNUtd1J7OapklysnzbhfF4nOXl5U5mAQAAANBfSsUAAAAAAADbyHw+Lx0BAAAAgC1IqRgAAAAAAKCwLjcBTyaTJMl0Ou1sJnBpdbm9vCvredZfY/rCNnUAAIDuKBUDAAAAAAAA9Ehd1/niE1/Jlbv3lI5y2lp2JEm+8szRwknOOPrMgdIRAAAAthWlYgAAAAAAAICeuXL3nrzpjl8pHaPX9t/3sdIRAAAAtpWF0gEAAAAAAAAAAAAAgLJsKgYAAAAAAAAAAIAtoGmaHFl9IXd/4ZOlo/Ta/tWv5+rm+dIxYMtRKgYAAAAAAAAAAIAt4ujxo9m/+vVNzTj20rGcaE90lKhbC9VCrrjsik3NOHr8aK7ONR0lguFQKgYAAAAAAAAAAOA1m81mqeu6dIxzrOeZTCaFk5xrPB5neXl503Pe9ra3dfKcN02T+Xy+6TkXw+LiYkaj0abnjMfjDtLAsCgVAwAAAAAAAAAA8JrVdZ19e7+SPd95U+kop+04fnLD7dqT/SnMHnj+YGezuigmA3wrSsUAAAAAAAAAAABckD3feVM++IO/VDpGr93zyMdLRwDYkIXSAQAAAAAAAAAAAACAspSKAQAAAAAAAAAAAGDglIoBAAAAAAAAAAAAYOAuLx0AAAAAAAAAAACgL2azWeq67mTW+pzJZNLJvPF4nOXl5U5mAcD5lIoBAAAAAAAAAAAugsXFxdIRAGDDlIoBAAAAYAvrcnNO0zRJktFo1Mk8m3MAYHvo8v1GV7re+NcV738AYHvw7zkAQ6VUDAAAAAAkSebzeekIAEAP1XWdx574q1y2q5sTj7pwor0sSfKlQy8UTnLGSytN6QgAAACwKUrFAAAAALCFdbk5Z33T33Q67WwmALA9XLZrlGtu//nSMXrthfs/UToCAADAttA0TY4cXs1dD3+2dJRe23/4UK6ujnY6c6HTaQAAAAAAAAAAAADAlmNTMQAAAAAAAAB0YDabpa7rTmatz1m/oshmjcfjTq90AgAAF8toNMpae2XufOuPlY7Sa3c9/NnsGO3qdKZSMQAAAAAAAAD0zOLiYukIAACkuxPHmqbJfD7vIFH3FhcXMxqNNj3HiWxbn1IxAAAAAAAAAHRAgQIAYPv5kz/5kzz7zEquvHzHpuYce+l4TrQnOkrVraPffDFHnnt+czOOr6VpGu+JtzilYgAAAAAA2Caapkm7upq1B+4rHaXX2pWVNMeOlY4BAAAAbBFXXr4jb7r2+tIxem3/6tOlI9ABpWIAAAAAAAAAAACAVzAajbJ24rl86EfeVzpKr939+d/LjtF1pWOwSUrFAAAAAACwTYxGo6xccUV2vPuO0lF6be2B+zJaWiodAwAAAAB6RamYS2o2m6Wu605mNU2T5OSH5F0Yj8dZXl7uZBYAAAAAAAAAAADAVqJUzJY1n89LRwAAAAAAAAAAAADYFpSKuaS63AQ8mUySJNPptLOZAAAAAAAAAAAAAEOkVAwAAAAAAMCWcmLlqczv/3TpGKedOPxskmRh5+sLJznjxMpTydLO0jEAAACALUSpGAAAAAAA4ALNZrPUdV06xjnW86xf8a8vxuNxJ1c0HI/HHaTpVr36TJJk3KcS79LOXj5XAAAAW9GB1adz9+d/r3SM054+8o0kyfVXv65wkjMOrD6dW3Jd6RhsklIxAAAAAADABarrOo898eUs7FoqHeW0E22VJPnioWcLJznjxMqhzmZ1UUzu2nqBezqdFk4CAABA1/p4wuZavZIk2XFTf0q8t+S6Xj5XvDZKxQAAAADAIDRNk3b1cI7f/1DpKL3WrjyX5lhbOgZsKQu7lnLlu3+8dIxeO/rAvaUjAAAAwAVxcitDslA6AAAAAAAAAAAAAABQlk3F0FOz2Sx1XZeOcY71POtnuvTFeDzu5RlBAAAAQL+MRqOsXFHl8tvfXjpKrx2//6GMlm4oHQMAAAAAgEtMqRh6qq7rfGXvF7Nn51Wlo5y246W1JMnRr+0rnOSMA4dfLB0BgG2qyxN8mqZJcrLE0gUn1AAAAAAAUIrPzwEAti+lYuixPTuvyq++9XtKx+i133j4q6UjAMCrms/npSMAAAAAAEDv+PwcAKBflIoBAOAVdLnJYDKZJEmm02lnMwEAuDDtynM5fv9DpWOc1h5+IUlS7bymcJIz2pXnkqUbSscAAAB6yufnAHBhutz2vz5n/d/SzbLtn3VKxQAAAADAIIzH49IRXqZePfnh/7hPJd6lG3r5XAEAAJvTZZGpK10XorqiWAVA3y0uLpaOwDalVAwAAPRClx9oN02TJBmNRp3M8wEyAGwPffz33FYuAADgUqnrOl9+fF+uf/2bSkc5baHdkSR57uljhZOc8fSz+0tHAGCb6uPnk3A+pWIAAGDbmc/npSMAAAAAAPTO9a9/U37ytjtLx+i13/+ju0pHAAAoRqkYAADohS7PzLXxDwAAAADgXE3T5PnD31SafRVPP7s/33zpO0rHAAAoQqkYAAAAAOA1mM1mqeu6k1nrc9ZPitqs8XjsMooAAGwJTdNkdfWbue+Bj5aO0msrK/tz7JiCKwAAl4ZSMQAAAABAIYuLi6UjAAAAAzEajfLcZcfyk7fdWTpKr/3+H92V666/onQMAIAilIoBAAAAAF4Dm4ABAGDzRqNRrrjiWO5496+VjtJr9z3w0SwtKbgCAHBpLJQOAAAAAAAAAAAAAACUZVMxAAAAAAAAAAAAAL1x4PCh3PXwZ0vHOO3pI88lSa6/+rrCSc44cPhQbrlxV6czlYoBAAAAAAAAAAAA6IXxeFw6wsus1SdLxTs6LvFuxi037ur8uVIqBgAAAAAAAAAAAKAXlpeXS0d4mclkkiSZTqeFk1xcC6UDAAAAAAAAAAAAAABlbahUXFXVbVVVfbmqqn1VVX3gFX6+p6qqh6qq+vOqqh6rqupdZ/3sV0897stVVf39LsMDAAAAAAAAAAAAAJt3+asdUFXVZUl+O8k7kxxM8khVVfe3bbv3rMPuTHJv27azqqrenOTBJDefuv2eJN+fZJTkv1dV9b1t277U9S8CAAAAAAAAAAAAAFyYjWwq/qEk+9q2/WrbtmtJPpPkjvOOaZNce+r2ziTNqdt3JPlM27ZH27b96yT7Ts0DAAAAAAAAAAAAAHpiI6XiG5M8edb3B0/dd7aPJHlfVVUHc3JL8S++hsemqqr3V1X1aFVVjx46dGiD0QEAAAAAAAAAAACALmykVLwR703yqbZtb0ryriS/W1XVhme3bfvJtm3f0rbtW5aWljqKBAAAAAAAAAAAAABsxOUbOOZrSd541vc3nbrvbD+T5LYkadv2T6uquirJ7g0+FgAAAAAAAAAAAAAoaCOl4keS3FpV1XfnZCH4PUl+4rxjDiR5R5JPVVX1fUmuSnIoyf1J/nNVVb+VZJTk1iT/p6PsAL00m81S13Uns5qmSZKMRqNO5o3H4ywvL3cyCwAAAAAAAAAAgO3jVUvFbdser6rqF5J8LsllSf5T27Zfqqrq15M82rbt/Ul+Ocl/qKrql5K0SX66bds2yZeqqro3yd4kx5P8s7ZtX7pYvwzAZiwvL+epp57a9JyjR4/mxIkTHSTK6TnPPvtsJ/P27t2bP/7jP970nDe84Q2ZzWYdJAIAAAAAAM7XNE2Orh7J/vs+VjpKrx195kCatatLxwCAQWuaJkeeP5J7Hvl46Si9tv/5g7m68b4F6L+NbCpO27YPJnnwvPs+fNbtvUl+9Fs89u4kd28iI8Alcfjw4cyPHMmODb0yfmuXJ8lCF4mSY+3Jr1csdFNSTk7kpaPHNzVh7fjJ5woAAAAAAAAAAIDtY5PVOYDtYzQa5br22fzzt15ZOkqv/buHj+Y7RqPSMQAAAAAAYNsajUY5suNo3nTHr5SO0mv77/tYRrv9XQcAShqNRll7aZ4P/uAvlY7Sa/c88vHsGC2WjgHwqpSKAQAAAAAALlDTNDmxupqjD9xbOkpoTU+sAAAgAElEQVSvnVj5mzTHXiwd4xyz2Sx1XXcya33OZDLpZN54PM7y8nInswAAAAA2SqkYAAAAAAAANmFxcXtvHGuaJi+tvpAX7v9E6Si99tJKk+bYNaVjAAAAwAVTKgYAAAB6p8utcV3pevtcV2yxA4CyRqNRnr3iqlz57h8vHaXXjj5wb0ZLry8d4xzeQwGUt7KyP/c98NHSMU47fPipJMnOnW8onOSMlZX9WVq6pXQMAAAGQqkYAADYFMW/jVP8g42r6zqPPfHFZPeVpaOcZS1J8tgzXymc4yzPHC2dAOCS6PI9Z9M0SU4WQbvgPR4wBKPRKN+44oVcc/vPl47Say/c/4mMlmwqho0aj8elI7zM6urJ//ZfWrqicJIzlpZu6eVzBQDA9qRUDAC8Zv6YC5ytruvsffyxXNujhVPH25NfDz79WNkgZ1l9tnQC2IJ2X5nL/tFNpVP02kt/eLB0BIAtZz6fl44AAED6uTF+fUnBdDotnAQAAMpQKgYAivLHXLpma+7GdVnCv/b1yY/cVnUya7v6/B+1pSMAAGxZXRZOFEUAAAAAAF6ZUjEA8Jr5Yy59Vtd1Hn/8sbzudaWTnHHixMmvTz3Vn6253/hG6QQAAAAAAAAAQJ8oFQMAsO287nXJO95ZOkW//Y//VjoBAABwsbQrK1l74L7SMU5rDx9OklQ7dxZOcka7spIsLZWOAQAAAAC9olQMAAAAAADbxHg8Lh3hZerV1STJuE8l3qWlXj5XAAAAAFCSUjEAAAAAAGwTy8vLpSO8zGQySZJMp9PCSQAAAACAb0epmA2ZzWap67p0jHOs51n/QLovxuNxLz+4BwCA83X5Pr9pmiTJaDTqZJ731QAAAAAAAACXllIxG1LXdfY9vjd7dl5bOsppO146niRZaw4WTnLGgcOrpSMAAEAR8/m8dAQAAAAAAAAANkGpmA3bs/Pa3Pl3f6R0jF676399vnQEAADYsC43AbukNQAAQ3Zi5VCOPnBv6RinnTj8XJJkYed1hZOccWLlULL0+tIxAAAAAPg2lIoBAAAAAAAu0Hg8Lh3hZerVbyRJxn0q8S69vpfPFQAAAABnKBVDTzVNkyOHX8xvPPzV0lF67cDhF3N11ZSOAQAAAAAMVJdXAOmKK4kAAAAAJLPZLHVddzJrfc765y6bNR6Pe/m5klIxAAAAAAAAAAAAAHwLi4uLpSNcEkrF0FOj0ShH22/mV9/6PaWj9NpvPPzVXDkalY4BAAAAAAAAAABAj/RxE3DfKRUDAADAJdLlJZaapkly8oTELvT1EksAAAAAAPTbgecP5p5HPl46xmlPf/NQkuT671gqnOSMA88fzC25tXQMgFelVAzAlqecAwAM0Xw+Lx0BAAAAAICBG4/HpSO8zFp9LEmy442LhZOccUtu7eVzBXA+pWIAOItyDgBwMXV5stFkMkmSTKfTzmYCAAAAAMBr0cclWz4/B7hwSsUAbHnKOQAAAAAAAAAAAJuzUDoAAAAAAAAA8P/Zu9sYua77PODPWZFyNpJFvZAiPZLlCCMFbhGkNSo4jQs3qQMjBhpHyKfQKVAkH+pikxiFgflgoEEapCpSBFu4LeIO4DZ1kKKwHBhxJbmqFFWqYadyALmAS9uS7GiMSqYnlvkiLkVGFt9OP5BLLl9MLrmzPHf3/n7AYnjvnDl6eCEMyd3n/gcAAACgLZOKAQAANpjxeJzJZNI6xnmW8yxP/e+K4XDYyY9eAwAAAGA2Zvm9sll/j8v3pgAA2GiUigEAADaYyWSSb7y4Jzfd3jrJOcfr6cf/9/09bYOscPRg6wQAAAAAbCTz8/OtI6y7Vw++nP/6xEOtY5z12uvfS5Lc9tZdjZOc8+rBl3PrzvtaxwAAaEKpGAAAYAO66fbkJ36htI7RaV//fG0dAQAAVs2URQC4Nv6MWr3hcNg6wkUOHDmWJLl159bGSc65ded9nbxWAADXg1IxAAAAAADAJtKHKYsAwNXrYgF7+SaoxcXFxkkAAEiUigEAgDWaTqc5vJQ8+4SpsJdz+GAyPTltHQMAAOioLpZ8AAAAAOiXudYBAAAAAAAAAAAAAIC2lIoBAIA1GQwGrSNc5Ojrp7+6povXCgAAAAAAAACSZEvrAADA9TMejzOZTFrHOM9yntFo1DjJ+YbDoY8dhVUaDoetI1xkcuT0e8vdOzuUbWc3rxUAAAAAAAAAJErFANArk8kk33xhT3bc2jrJOeXU6ceDf7WnbZAV9h1qnQA2llkV8Lt448MyNxoAAAAA19ub+1/Jy4/8fusYZx1b+n6S5MZtdzZOcs6b+19Jtt/fOgYAAMCmoVQMHfbK0g/ye3/+7dYxzvr+0WNJkjtvurFxknNeWfpB7r+rdQrYWHbcmvzy+/wV4HI+88yJ1hGANZqfn28dAQAAAOCadfHTjiZLp39ONNz+lsZJVth+fyevFQAAwEalUQQd1cVvgBw7MznwLXd1J9v9d3XzWgEAV88kYAAAAIDTuvh9ktFolCRZXFxsnAQAAID1olQMHeWbRQAAAAAAAAAAAMD1Mtc6AAAAAAAAAAAAAADQlknFAAAbwHg8zmQymcle0+k0STIYDGay33A47NSE/el0mqWl5OmnWifpttdeS06dmraOwTWaTqc5upR8/fO1dZROO3ogmZ7w/zkAAAAAAADAaigVAwD0zBtvvNE6AgAAAAAAAAAAHaNUDACwAcxyEvBoNEqSLC4uzmzPLhkMBpmb25+fe3/rJN329FPJrl2zmVbN9TcYDHJsy/78xC+U1lE67eufrxnc6f9zAAAAAAAAgNVQKgYAAAAAAAAAAKCZ8XicyWQyk72W91ketrRWw+FwpkOgALpMqRgAAAAAAAAAAIBNYX5+vnUEgA1LqRgAAAAAAAAAAIBmTAIG6Ia51gEAAAAAAAAAAAAAgLZMKgYAAACABsbjcSaTSesY51nOMxqNGic533A4NK0GAAAAAADWmVIxAAAAADQwmUyy58XnU+64pXWUs2o9kST52r69jZOcUw8cbh0BAAAAAAB6QakYAAAAABopd9ySLQ++p3WMTjvxyLOtIwAAAAAAQC8oFQMAAAAAsC7G43Emk0nrGOdZzjMajRonOd9wOMzCwkLrGAAAAABAjykVAwAAAACwLiaTSfa8+ELKHbe3jnJWrTVJ8rV9rzZOck49cLB1BAAAAAAApWLY7GY5DWbWU1xMXwEAAADY/Modt2frBz/QOkanHX/sidYRAAAAAACUioHVm5+fbx0BAAAAAAAAAAAAWAdKxbDJmQQMAAAAAAAAAAAAXIlSMQAAAAAAcJ7xeJzJZDKTvZb3GY1GM9lvOBwapgANnDwwzZFH/0PrGGedWtqfJJnbtr1xknNOHpgmO368dQwAAAC4ZkrFADQzyx9Ozcqsf8g1K35YBgAAAGxU8/PzrSMAazQcDltHuMjk8KtJkuGOmxsnWWHHj3fyWgEAAMBqKRUD0MxkMsm3XtiTXdtK6yhnzZ2sSZLD0681TnLO95Zq6wgA0Htuhlo9N0MBwObgz3NgpS6+Jyz/W2hxcbFxEgAAANg8lIoBaGrXtpJf+9mtrWN02qe+cLx1BADovclkkq+9uCdbu/Opujlx5vHF/Xua5ljp+P7WCQAAAAAAAIBrpVTMqkyn0xw9dDgPffHZ1lE67eVDh3NTpq1jAPxQ0+k0h5eSzzxz4sqLe+z7h5IfVO/nAJxv6/Zkx4NzrWN02r5HTrWOAAAAAAAAAFwjPw0FAAAAAAAAAAAAgJ4zqZhVGQwGOZZT+a2//57WUTrtoS8+mxsHg9YxAH6owWCQHyn788vv81eAy/nMMydy+9u8nwMAXIvxeJzJZDKTvabT058eMZjRv7WHw2EWFhZmshcAAAAAAMBmo1EEAAAAQCe98cYbrSMAAAAAAAD0hlIxAAAAADMzy0nAo9EoSbK4uDizPQEAAAAAALi0udYBAAAAAAAAAAAAAIC2lIoBAAAAAAAAAAAAoOeUigEAAAAAAAAAAACg57a0DgAAAABAe+PxOJPJpHWM8yznGY1GjZOcbzgcZmFhoXUMAAAAAACAmVIqBgAAACCTySR7XvxGsv3m1lFWOJ4k2bP/5cY5Vth/pHUCAAAAAACAdaFUDAAAAMBp22/Olgf/VusUnXbikf/bOgIAAKzaLD+RZNafJOITQAAAALpHqRgAAAAAAACAy5qfn28dAegINywAAGxeSsUAAAAAAAAAm5BiHdB1blgAAOiWVZWKSykfSPLvktyQ5D/VWv/1Bc9/PMk/OHP4o0nurLXeeua530/yD5PMJXkqyT+rtdbZxAcAAAAAAAAA4HpxwwIAwOZ1xVJxKeWGJJ9I8v4ke5M8V0p5tNb6/PKaWutHV6z/SJJ3nfn1e5L8vSQ/eebpP0/yM0m+MKP8AMBV2nco+cwzJ1rHOOvQkdOPt97cNsdK+w4lt7+tdQoAAAAAAAAAALh+VjOp+N1JXqq1fjtJSikPJ3kwyfM/ZP2HkvyLM7+uSX4kyY1JSpKtSV5dS2AA4NoNh8PWES7y2mSSJLn9bd3JdvvbunmtAADW03Q6TQ4s5cQf/u+1bXTiVNLVD6kqJdkyt7Y9jp/M9Nh0NnkAAAAAAAA6ZDWl4ruSfGfF8d4kP3WphaWUdyS5N8kzSVJr/XIp5X8l+aucLhX/Qa31hUu87sNJPpwk99xzz9XkBwCuQhc/jmo0GiVJFhcXGycBAOi3bdu25Y033ljzPm+eejOnTp2aQaLZm5uby1u2vmVtm2w9fa0AAAAAAAA2m9WUiq/G7iSfrbWeTJJSyn1J/kaSu888/1Qp5b211i+tfFGt9ZNJPpkkDzzwQEdH2QAAAABsXuPxuHUEAAAAAAAAGlpNqfi7Sd6+4vjuM+cuZXeS31hx/EtJ/qLWeiRJSin/I8lPJ/nSJV4L0Nx3l07l3//5m61jnLXv6OnpXjtuWuPH887Qd5dO5f67WqcAAAAAAAAAAABgllZTKn4uyf2llHtzuky8O8mvXLiolPLOJLcl+fKK068k+SellN9LUpL8TJJ/u9bQAOthOBy2jnCRE5NJkuRH7+pOtvvv6ua1AljptdeSp59qneKc118//fjWt7bNsdJrryW7drVOAQAAAAAAAAB0xRVLxbXWE6WU30zyZJIbkvznWus3Sim/m+QrtdZHzyzdneThWmtd8fLPJnlfkq8lqUmeqLU+NtPfAcCMLCwstI5wkdFolCRZXFxsnARg4+jijQ9Hj56+SWTXru5k27Wrm9cKAAAAAAAAAGhjNZOKU2t9PMnjF5z77QuOf+cSrzuZ5J+uIR8AAFwVN4kAAAAAAAAAAFy9VZWKAQAAAK6n6XSaHH4zJz+3t3WUbtv/ZqbHpq1TAAAAAAAAsAnMtQ4AAAAAAAAAAAAAALRlUjEAAADQOYPBIPtvPJobfunu1lE67eTn9mawfdA6BgAAAAAAAJuAUjEAAMAGdPRg8vXP19YxzvrB4dOPP3JL2xwrHT2Y5M7WKQAAAAAAAAA2BqViAACADWY4HLaOcJHJ65MkyY/d2aFsd3bzWgEAAAAAAAB0kVIxq/bK0uE89MVnW8c469UjR5MkO2++qXGSc15ZOpz7fOos0APj8TiTyWQmey3vMxqNZrLfcDjMwsLCTPYC6Kouvs8tv48vLi42TgIAAAAAAADAtVAqZlW6ON3r2JkS2o2DuxsnOee+QTevFUCXzc/Pt44AAAAAAAAAAAC9p1TMqpiEBqyH6XSa1w/VfOoLx1tH6bS/OlRzJNPWMc7TxT8XAAAAAAAAAACAa9frUvEsP7p9Oj1d9hoMBjPZz0e3AwAAAAAAAAAAAHC99LpUPEtvvPFG6wgAG85gMMjhHMiv/ezW1lE67VNfOJ5bZnTTCtffLG9impXlPMtT/7vCTVUAAAAAAAAAAO30ulQ8y9LKcilncXFxZnsCABvfZDLJiy/syR23tU5yTj11+nHf9/a0DbLCgddaJwC4vOl0muOHk32PnGodpdOO70+mx6atYwDQIdPpNPXwUo4/9kTrKJ1WDxzM9PjJ1jEAAAAAgJ7rdakYAOB6uOO25IM/N9c6Rqc99rSSHgAAAAAAAABAS0rFAAAAcAWDwSCHb9yfHQ+6SeRy9j1yKoPtg9YxAOiQwWCQA1tvyNYPfqB1lE47/tgTGezY2ToGAAAAANBzfhoKAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPbeldQAAAAAAADaveuBgjj/2ROsYZ9Wl15MkZdtbGyc5px44mOzY2ToGAAAAANBzSsUAAAAAAKyL4XDYOsJFJoePJEmGXSrx7tjZyWsFAAAAAPSLUjEAAAAAAOtiYWGhdYSLjEajJMni4mLjJAAAAAAA3TLXOgAAAAAAAAAAAAAA0JZSMQAAAAAAAAAAAAD0nFIxAAAAAAAAAAAAAPTcltYBAAAAAADgcsbjcSaTyUz2Wt5nNBrNZL/hcJiFhYWZ7AUAAAAA0JJSMQAAAAAAvTE/P986AgAAAABAJykVAwAAAADQaSYBA2wOJs8DAABAtykVAwAAAAAAABuKyfMAAAAwe0rFAAAAAAAAwLozCRgAAAC6ba51AAAAAAAAAAAAAACgLaViAAAAAAAAAAAAAOg5pWIAAAAAAAAAAAAA6DmlYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOi5La0DXIvxeJzJZNI6xnmW84xGo8ZJzjccDrOwsNA6BgAAAAAAAAAAAAAdtiFLxZPJJC+98ELese321lHOuvFkTZIcn77aOMk5Ly8dbB3hIrMshM+6yK2ADQAAAAAAAAAAAPTVhiwVJ8k7tt2e33rvz7eO0WkPfenJ1hHW1fz8fOsIAAAAAAAAAAAAAJvChi0VszGZBAwAAAAAAAAAAADQPUrFAABwCePxOJPJZCZ7Le8zGo1mst9wOHSzFgAAAAAAAAAwU0rFAACwzubn51tHAAAAAAAAAAC4LKViAAC4BJOAAQAAAAAAAIA+mWsdAAAAAAAAAAAAAABoS6kYAAAAAAAAAAAAAHpOqRgAAAAAAAAAAAAAek6pGAAAAAAAAAAAAAB6TqkYAAAAAAAAAAAAAHpuS+sAAAAAAJe0/82c/Nze1inOWTp2+nHbjW1zrLT/zWR76xAAAAAAAABsBkrFAAAAQOcMh8PWES4yWZokSYbbO5RtezevFQAAAAAAABuPUjEAAADQOQsLC60jXGQ0GiVJFhcXGycBAAAAAACA2ZtrHQAAAAAAAAAAAAAAaMukYgAAgJ4aj8eZTCYz2Wt5n+VJrms1HA47OakWAAAAAAAAYLNSKgYAAGDN5ufnW0cAAAAAAAAAYA2UigEAAHrKJGAAAAAAAAAAls21DgAAAAAAAAAAAAAAtKVUDAAAAAAAAAAAAAA9t6V1AAAAAADoo+l0mnr4cE488mzrKJ1WDxzO9Pi0dQwAAAAAANj0TCoGAAAAAAAAAAAAgJ4zqRgAAAAAGhgMBjmw9VS2PPie1lE67cQjz2awY9A6BgAAAAAAbHobslQ8nU7z14eW8tCXnmwdpdNePnQwP5qTrWMAAAAA8EPUA4dz4pFnW8c4qy4dTZKUbTc1TnJOPXA42dE6BQAAAAAAbH4bslQMAAAAABvdcDhsHeEik8OTJMlwx92Nk6ywo5vXCgAAAAAANpsNWSoeDAY5nhvyW+/9+dZROu2hLz2ZrYOdrWMAAAAAcAkLCwutI1xkNBolSRYXFxsnAQAAAAAArre51gEAAAAAAAAAAAAAgLY25KRiAICNYjqd5vBS8tjTp1pH6bQDryXHT01bxwC4rOP7k32PdOf9/MTS6cct29rmWOn4/iTbW6eA/hmPx5lMJjPZa3mf5YnFazUcDjs5kRkAAAAAALiYUjEAAABcwXA4bB3hIpOl08W/4fYOZdvezWsFrN78/HzrCAAAAAAAQCNKxQAA62gwGGTr3P588OfmWkfptMeePpUduwatYwD8UF2csrk8RXRxcbFxEqC1Lr5HAQAAAAAAG492CwAAAAAAAAAAAAD0nFIxAAAAAAAAAAAAAPScUjEAAAAAAAAAAAAA9NyW1gEAADa7A68ljz19qnWMs5ZeP/247a1tc6x04LVkx67WKQAAAAAAAAAA+kupGABgHQ2Hw9YRLnL46CRJsmNXd7Lt2NXNawUAAAAAAAAA0BdKxQAA62hhYaF1hIuMRqMkyeLiYuMkAAAAAAAAAAB0xdxqFpVSPlBK+WYp5aVSyscu8fzHSylfPfP1rVLKoRXP3VNK+bNSygullOdLKT82u/gAAAAAAAAAAAAAwFpdcVJxKeWGJJ9I8v4ke5M8V0p5tNb6/PKaWutHV6z/SJJ3rdjij5P8q1rrU6WUm5OcmlV4AAAAAAAAAAAAAGDtVjOp+N1JXqq1frvWeizJw0kevMz6DyX5dJKUUv5mki211qeSpNZ6pNb612vMDAAAAAAAAAAAAADM0BUnFSe5K8l3VhzvTfJTl1pYSnlHknuTPHPm1I8nOVRK+dMz5/9nko/VWk9e8LoPJ/lwktxzzz1Xkx8AoBfG43Emk8lM9lreZzQazWS/4XCYhYWFmewFAAAAAAAAAEAbqykVX43dST67ojS8Jcl7k7wryStJPpPkV5P84coX1Vo/meSTSfLAAw/UGWcCuK4U/4Cum5+fbx0BAAAAAAAAAICOWU2p+LtJ3r7i+O4z5y5ld5LfWHG8N8lXa63fTpJSyn9L8ndzQakYgEtT/AOWuSEAAAAAAAAAAID1tJpS8XNJ7i+l3JvTZeLdSX7lwkWllHcmuS3Jly947a2llB211n1J3pfkK2tODdBhin8AAAAAAAAAAABsNHNXWlBrPZHkN5M8meSFJH9Sa/1GKeV3Sym/uGLp7iQP11rriteeTDJK8nQp5WtJSpL/OMvfAAAAAAAAAAAAAACwNquZVJxa6+NJHr/g3G9fcPw7P+S1TyX5yWvMBwAAAAAAAAAAAACssytOKgYAAAAAAAAAAAAANjelYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOg5pWIAAAAAAAAAAAAA6DmlYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOg5pWIAAAAAAAAAAAAA6DmlYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOi5La0DAAAAAKyX8XicyWQyk72W9xmNRjPZbzgcZmFhYSZ7AQAAAAAAwFopFQMAAACswvz8fOsIAAAAAAAAsG6UigEAAIBNyyRgAAAAAAAAWJ251gEAAAAAAAAAAAAAgLaUigEAAAAAAAAAAACg55SKAQAAAAAAAAAAAKDnlIoBAAAAAAAAAAAAoOeUigEAAAAAAAAAAACg55SKAQAAAAAAAAAAAKDnlIoBAAAAAAAAAAAAoOeUigEAAAAAAAAAAACg55SKAQAAAAAAAAAAAKDntrQOAEC/fW+p5lNfON46xlkHjtQkyR03l8ZJzvneUs0tg9YpAAAAAAAAAACAzUypGIBmhsNh6wgX2TeZJEluGXQn2y2Dbl4rAAAAAAAAAABg81AqBqCZhYWF1hEuMhqNkiSLi4uNkwAAAAAAAAAAAFw/c60DAAAAAAAAAAAAAABtKRUDAAAAAAAAAAAAQM8pFQMAAAAAAAAAAABAzykVAwAAAAAAAAAAAEDPKRUDAAAAAAAAAAAAQM8pFQMAAAAAAAAAAABAz21pHQAAAAD6YjweZzKZzGSv5X1Go9FM9hsOh1lYWJjJXgAAAAAAAMDGo1QMAAAAG9D8/HzrCAAAAAAAAMAmolQMAAAA14lJwAAAAAAAAEBXzbUOAAAAAAAAAAAAAAC0pVQMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPbeldYBr9fLSwTz0pSdbxzjre0deT5LsuvmtjZOc8/LSwdw32Nk6BgAAAAAAAAAAAAAdtyFLxcPhsHWEixybHEmSbO1Qife+wc5OXisAAAAAAAAAAAAAumVDlooXFhZaR7jIaDRKkiwuLjZOAgAAAAAAAAAAAABXZ651AAAAAAAAAAAAAACgLaViAAAAAAAAAAAAAOg5pWIAAAAAAAAAAAAA6DmlYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOg5pWIAAAAAAAAAAAAA6DmlYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOg5pWIAAAAAAAAAAAAA6DmlYgAAAAAAAAAAAADoOaViAAAAAAAAAAAAAOi5VZWKSykfKKV8s5TyUinlY5d4/uOllK+e+fpWKeXQBc/fUkrZW0r5g1kFBwAAAAAAAAAAAABmY8uVFpRSbkjyiSTvT7I3yXOllEdrrc8vr6m1fnTF+o8kedcF2/zLJF+cSWIAAAAAAAAAAAAAYKZWM6n43UleqrV+u9Z6LMnDSR68zPoPJfn08kEp5e8k2Znkz9YSFAAAAAAAAAAAAABYH6spFd+V5DsrjveeOXeRUso7ktyb5Jkzx3NJ/k2S0eX+A6WUD5dSvlJK+cq+fftWkxsAAAAAAAAAAAAAmJHVlIqvxu4kn621njxz/OtJHq+17r3ci2qtn6y1PlBrfWDHjh0zjgQAAAAAAAAAAAAAXM6WVaz5bpK3rzi++8y5S9md5DdWHP90kveWUn49yc1JbiylHKm1frChPysAAB7cSURBVOxawgIAAAAAAAAAAAAAs7eaUvFzSe4vpdyb02Xi3Ul+5cJFpZR3JrktyZeXz9Va/9GK5381yQMKxQAAAAAAAAAAAADQLXNXWlBrPZHkN5M8meSFJH9Sa/1GKeV3Sym/uGLp7iQP11rr+kQFAAAAAAAAAAAAANbDaiYVp9b6eJLHLzj32xcc/84V9vijJH90VekAAAAAAAAAAAAAgHV3xUnFAAAAAAAAAAAAAMDmplQMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD2nVAwAAAAAAAAAAAAAPadUDAAAAAAAAAAAAAA9p1QMAAAAAAAAAAAAAD23pXUAAFir8XicyWQyk72W9xmNRjPZbzgcZmFhYSZ7AQAAAAAAAAAArBelYgBYYX5+vnUEAAAAAAAAAACA606pGIANzyRgAAAAAAAAAACAtZlrHQAAAAAAAAAAAAAAaEupGAAAAAAAAAAAAAB6TqkYAAAAAAAAAAAAAHpOqRgAAAAAAAAAAAAAek6pGAAAAAAAAAAAAAB6TqkYAAAAAAAAAAAAAHpOqRgAAAAAAAAAAAAAek6pGAAAAAAAAAAAAAB6TqkYAAAAAAAAAAAAAHpOqRgAAAAAAAAAAAAAek6pGAAAAAAAAAAAAAB6TqkYAAAAAAAAAAAAAHpOqRgAAAAAAAAAAAAAek6pGAAAAAAAAAAAAAB6bkvrAC2Nx+NMJpOZ7LW8z2g0msl+w+EwCwsLM9kLAAAAAAAAAAAAAC6n16XiWZqfn28dAQAAAAAAAAAAAACuSa9LxSYBAwAAAAAAAAAAAEAy1zoAAAAAAAAAAAAAANCWUjEAAAAAAAAAAAAA9JxSMQAAAAAAAAAAAAD0nFIxAAAAAAAAAAAAAPScUjEAAAAAAAAAAAAA9JxSMQAAAAAAAAAAAAD0nFIxAAAAAAAAAAAAAPScUjEAAAAAAAAAAAAA9JxSMQAAAAAAAAAAAAD0nFIxAAAAAAAAAAAAAPScUjEAAAAAAAAAAAAA9JxSMQAAAAAAAAAAAAD0nFIxAAAAAAAAAAAAAPTcqkrFpZQPlFK+WUp5qZTysUs8//FSylfPfH2rlHLozPm/XUr5cinlG6WUPaWUX571bwAAAAAAAAAAAAAAWJstV1pQSrkhySeSvD/J3iTPlVIerbU+v7ym1vrRFes/kuRdZw7/Osk/rrX+ZSllkOT/lFKerLUemuVvAgAAAAAAAAAAAAC4dquZVPzuJC/VWr9daz2W5OEkD15m/YeSfDpJaq3fqrX+5ZlfT5N8P8mOtUUGAAAAAAAAAAAAAGZpNaXiu5J8Z8Xx3jPnLlJKeUeSe5M8c4nn3p3kxiSTSzz34VLKV0opX9m3b99qcgMAAAAAAAAAAAAAM7KaUvHV2J3ks7XWkytPllLeluS/JPm1WuupC19Ua/1krfWBWusDO3YYZAwAAAAAAAAAAAAA19NqSsXfTfL2Fcd3nzl3KbuTfHrliVLKLUn+e5J/Xmv9i2sJCQAAAAAAAAAAAACsn9WUip9Lcn8p5d5Syo05XRx+9MJFpZR3JrktyZdXnLsxyeeS/HGt9bOziQwAAPz/9s49eK6yvsPP1wDhEpGYRIxAk6BYC6KAgaIi4xWUAhrFDhY1KVAHa6rWcaoMrTIqY71WLSpaBcQbKmqNqCBEELUNGAIJxBAIGAaRcpOL0QgCb/94382+u7+z57Zn3738Ps/Mzp7beznPOfvd93337DlCCCGEEEIIIYQQQgghhBBCCCGEEE1SeFGxc+4RYDlwMbAB+KZzbr2Zvc/Mjo02PR443znnomV/CxwOLDOza8PrgAbrL4QQQgghhBBCCCGEEEIIIYQQQgghhBBCCCGE6BPrvAZ4+CxevNitXr162NUQQgghhBBCCCGEEEIIIYQQQgghhBBCCCGEEGKsMbOrnXOLy2xbeKdiIYQQQgghhBBCCCGEEEIIIYQQQgghhBBCCCHEZDNydyo2s7uBW4ddj5rMBe4ZdiWmGXKeHjlPj5ynR87TI+fpkfP0yHl65Dw9cp4eOU+PnKdHztMj5+mR8/TIeXrkPD1ynh45T4+cp0fO0yPn6ZHz9Mh5euQ8PXKeHjlPj5ynR87TM67OFzjn5pXZcOQuKh5nzGx12VtEi2aQ8/TIeXrkPD1ynh45T4+cp0fO0yPn6ZHz9Mh5euQ8PXKeHjlPj5ynR87TI+fpkfP0yHl65Dw9cp4eOU+PnKdHztMj5+mR8/TIeXrkPD1ynp7p4Pxxw66AEEIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBBiuOiiYiGEEEIIIYQQQgghhBBCCCGEEEIIIYQQQgghpjm6qLhZPj/sCkxD5Dw9cp4eOU+PnKdHztMj5+mR8/TIeXrkPD1ynh45T4+cp0fO0yPn6ZHz9Mh5euQ8PXKeHjlPj5ynR87TI+fpkfP0yHl65Dw9cp4eOU+PnKdn4p2bc27YdRBCCCGEEEIIIYQQQgghhBBCCCGEEEIIIYQQQgwR3alYCCGEEEIIIYQQQgghhBBCCCGEEEIIIYQQQohpji4qFkIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBBimjNSFxWb2U5m9lMzmxHm55vZhWF6jpldZmZbzOzMrnSXm9lGM7s2vJ4Uls80s2+Y2SYzu9LMFpaow2Yzuy7kszpa/kQzu8TMbgrvs0vkdZGZ3d/ah2j5olCfTaF+O4Tly83sxBL5vtLM1rXqaGaHlUgTO1prZv9jZjNCmf8sz5n51vF8hpndZmZbupY/wczuihwdbGYXmtn+ZvZ1Od+WvpJzM9vZzH5gZjeY2Xoz+/doXezoKjNbFc75/cNynfNT832hmT0Q7e97SqQ518x+be3Ysjp4PtrMPi7PmfnW8bw8lOfMbG60fCcz+01Yt87MjjAfW+aZ2Uo535a+jvOvBkfXm9nZZrZ9WG5m9qnI+UFh+Twzu6hHXmPXvjH/GX5fiXwPieq31syWlEgTx41rzeyAKmXm5CvPnWl6xY3K53BOGXLemaaxuFGGCfe/0My2RnU8q0Sa083s9ijNUWH5/mZ2blH6nHzluTPNa823ux8zs8Vd604N+7XRzI4My3YwsyvMbLuivKN85LwzzUfM93fWmdl3zWy3aF0jzrvKq+v/OcHZJvMxz8LySm1CM1tgZmuCn/VmdkqJMj5qZi8usW+nRMf152a2b4k0vc6FUmXm5CvPnWnONj9mcn3X8r4+c115yXl7+73C/v4q1OVt0brGnHeVOcn+l5nZ3daO7SeXSNPrO6tUHzcnX3nuTNNrnDazbWA12o1y3rF92XHavpxXYRjHp+w+5R27nDQD6fP0ywR6Hnifp18m0HnSPk+/jLL/sO2jUZxYUWL7zHhvfY5b9cMEOj48fC4eMbPjutYtDfW6ycyWRssvzarnoJhA5+8w399ZZ/63sgXRupFwnkfB8eg5bm1mLw/xcpOZvbtH3mM3hjIoxtjzwMdQBsU4OrchjKEMihHxPzJjKINijD0PfAxlUIyjcxvBMZRMnHMj8wLeArwtmv8I8MowvQtwGHAKcGZXusuBxRn5/SNwVpg+HvhGiTpsBuZmLP8w8O4w/W7gQyXyeglwDHBh1/JvAseH6bOAN4fpnYFrSuQ7C7Aw/SzghhJptjmKPYcy75TnxjwfCswHtnQtPx/4ReTohsj5T4BXy3l15yHfF4XpHYCfAa/IcHQ2sKbLxUk656fk+8LuPEukORc4LkzHscXwseW18tyI5wOBhd37B3wG+FXwfSjwW9qx5cvAm+W8tvOjglcDvh6VeRTwo8j5lVGac4DnZ+Q1du2bsH/XADsX5LszsF2Yng/c1ZrPSXMuIW50LS9VZk6+8tyZplfcqHwOy3lp543FDflnIXB9RR+nA+/sse5S4C/kuRHPfwX8Zff+AfsCa4GZwCLgZmBGWPde4AQ5r+38CNrx6ENRmY05b8j/VfgYZ/iY1+qXVWoT4vt1M8P0rHAsnlJQxgLgxyX2bddo+ljgoj7OhVJlynNpz4cDB3V/PnrtFzXajXLesf184KAw/XjgRmDfpp1PI//LuutdIs3lZH9nlerjynNpH73GaXu2DajYbpTzKedvmXHavpyPw2eizD7lHbucNKczgD6PPE9JM/A+j5xPSZO0zzPJ/sN2Wyruz7LuukbrzqHmuJUcd2y/EP+b53lE4+LAE4FbwvvsMD07rFsKnCbntZ2/iNB3wf9O9o1Rc97H8cgctwZm4OPk3vhYvJbQz+vKu2xbcGTGUOR5SpqBj6HIecf2ycdQJtz/MkZkDEWep6QZ+BiKnE85f0dqDCXrNVJ3KgZOAL4Xzb8GuAjAOfcH59zPgT9VyO+VwJfC9AXAS8z8P+RqEOf1JeBVRQmccyuB38fLQvkvDvXpyMs590dgs5kdUpDvFhfOFnxD3uVtn8E2z6HMWcA9YV6e2/lW9uycW+WcuyNj1ZHAR8P0BcA+hHMbfyyehpxXdu6c+6Nz7rIw/TCwBtgzY7+eASyMHM0CnhTS6Zxvjji2OPxg3/ZhXp77wDl3jXNuc8aqVwOfdZ5VwFzg2rDuAmB/5LwWzrkfBq8OPzAWx5bzIue7mdn8sO6/8Z+DbsaufRP2+3Lg6LzEIQ4/EmZ3pHqbJM6rVJk5yHNnml5xo8453As570zTZNwow8T6HwDfx3e86yDPEc65Dc65jT3qcr5z7iHn3K+BTUDr+1mxpQ+ccz+O4tEqOmNLU85jKvsPMW3X0B92+B9KW22/Sm1C59zDzrmHwuxMwhO28spwzt0KzDGzJxfk/WA0W2csJc6rVJk5yHNnmiuA32WsavIzJ+ft7e9wzq0J078HNgB75O3XMNrq4+K/SRro48pzZ316jdPmtQ2qthvlvF2XsuO0/TqvwrCOT+E+9Tp2fTBIj0VMlOdEfZ5+mTTnqfs8/TKy/gfAsDxPlGPn3Gbn3Drgsa5VRwKXOOd+55y7D7gEeHlYtwJ4XdN1yWHSnF8W2tbQGVdGyXkeecej17j1IcAm59wtoS12Pv44dDDIvn1OXsn7ViUZS8+JxlAGxdg5H9IYyqAYuv8maWAMZVCMpedEYyiDYuycj+gYyhRG5qJi848y39uFix/MbBFwX9T5LOKccAvpf4tk7gHcBhBOkgeAOQX5OODHZna1mb0pWr579AH6P2D3kvXqZg5wf3TS/ob2lw7AauAFRZmY2RIzuwH4AVD2lu7nmNla/EVmt4Z8FgH34/91UDYPea5AOLdnAb8Mi/YCHsH/k6lMXeS8BOYfiXUMsDIs2gO4rRVbgPvwwX4R/iL655XMelr5B55r/rb/PzKz/UqWfYaZXQc8G7gD2jEceG7JPOS5IuHc3g3/T8eW8z8B80rWRc5LYGbbA2+g/UeQbY4y6jOlLmPevikbt//azNYD1wGnRMcqjzPMP3rsP8xsZtUyM+ogz+WpdA7n1EfOe6frK26ULGPi/QOLzOwa848sKutoeYgtZ1vn4wrluTd1PGeRd55fDxxcJhM5L+RE/J1/oCHnMX343yOUn1WXym1C848VXIffvw85535bUAb4Qbfnl8j7LWZ2M/5OIm8t2p7e50LpMjPqIM/laeIzJ+f56Rbin3BxZVjUiPOuMibeP/Ca0Aa5wMz2KrE9ZH9ngTznUcdzFnltA8WWTmo5tx7jtNCf8yoM+fiU7XNmHbsiGu3z9MsEe86i8fZ3HaaB84H2efplHPwDO5rZajNbZWaFN/oI9Ir3yWPLBDvOouc57vwFrzPNrGj8om+mgfOTKBFXUjrPo8zxsOxx67yY2c3YjKEMijH33IvG+/NNMgnOU4yhDIpR8s8IjKEMignwnEXy/nwVJsH5KIyh9GJkLirG313x/mh+PnB3ybQnOOf2x4t7Af5H/Loc5pw7CHgF8BYzO7x7A+eco49/YxVwF/CUoo2cc991zj0D/2+X95fIt+XoOPxxbzmaj7/wr7BM5LmM5yzm0vnv0/nAoyXrIuclMLPt8I8Z/5Rz7pau1VmxpVRdmH7+1wALnHPPBv4T/y/0Ik7F3wn6GPzF8u8Ky+fjL96W56nU8ZzFXLzzFvOBP5esi5yX5zPAFc65n9Wsyzi3b8rG7Sudc/vhB/JPNbMdC5K04sbB+MeMvStaVzY+dyPPzVDFv5z3pt+4UYZJ938H/vE9BwLvAL5mZrsWpPks8FTggJD+YxXLzEKeG8A59yjwsJk9vnBjOe+JmZ2Gb/t9tWjbis5j+vFfSNk2oXPuNufcs/BP9FlqZmX+nFY2tn/aOfdUfPvjX0vkm3cuDCO2FDKmngup+30ekPMMzGwW8G3g7a7zTj2tPPtxHjPp/r8PLAx5X0L7TiJ55H1nyXM2dTzXQbGlTS3nBeO0detSh2Een7Kxu+qxG0Sfp18m0XNl+mh/12FinSfq8/TLyPvHj4MvBv4O+ISZPbVg+7x4P4zYMomO65LK/8Q6N7PXA4vxj0Evw7C+T2MKj0cDvxWM0xjKoBhnz4U02J9vkrF2nnAMZVCMiv9RGUMZFOPsuQ6j4H+snY/QGEomo3RR8Vb8raZ7zffEOXd7eP898DXaj7+5HX9X2NaBeAJwb8m87gK+G+V1p4XHFIf3u8rULYN78Y883i7M7xnq2WJH/L6XwvnHG+xtZnMLtmuVcU/Iv7VfW/GPICosU56LPfdgK/6i4tY/ER4GtqftqGdd5Ly0888DNznnPhEtazlqxZKWo63ATmXqMt38O+cedM5tCdM/BLYvEVvuCI3kB5gaW0odc3ku9tyDrfjOyV7R/E5RfRRb2nWs5dzM3ou/8/M7osXbHGXUJ6su49y+qRq3NwBbgGcWbHeH8zwEnBPVpXKZEfJcnqrncC/kPIOG4kYZJtq/8482vTdMXw3cDDy9IM2dzrlHnXOPAf+FYstAPOeQd56D73N2PCazB3KegZktwz8u74TQ9u7Yr0Bd5zF1/d9O+9Fc3XWp3SZ0/o5m1+MHi/PKgOqf8/MpeKR8qEOvc6FOmS3kuTxNfZ/LeRfmn6zwbeCrzrnvRKsaa0NFTLR/59y9rn33ky8AzylRh17fWaXK7IE8lyevbaDY0s6vrvO8cdp+nVdhmMenap8zPnZ52w2iz9MvE+c5h0G0v+swkc4T9nn6ZeT9R+2MW/CPXj+wYPu8eD+M2DJxjnMoOsdT+Z9I52b2UuA04NjoHB8V53mUPh5d49ZF+xYzTmMog2JsPecwiP58k4yt88RjKINiJPyP0BjKoBhbzzkMoz9fhXF3PipjKJmMzEXFzj9SYkZ0RfiNwMKidGa2XevCoBDMj8Z3UgFWAEvD9HHAT5xzzsz2MLOVGXnt0vpXq5ntAhzRI6+lwPfCdoeY2XkV9tMBl4X6dOQVeHqrTDNbbmbLM+r5NDN/S3czOwjfcb43zN+Qsf02R/gPyM5Aa7sb8bfhv747Xa885Dnbc05d7sMPbJwYFu0HPBwNjGyrS1f5cl7CuZl9AB9E3961agWwNPifBVwe6ngjsACd81n+nxz5PwT/HdHyv9LMptzuv9VQxv/7ZxdgQ5i/EX9bfnmeWs/KnnvUpRVbloX8ngDMcO3HrCi2tOtZ59w+GTgSeF34sabFCuCN5jkUeCDP+bi2b7r3x8yWmNkHM/JeZOGCbzNbgL8D8eYwnxs3wjF5FZ3OMs/bIuS5XNyI6lL6HO6FnA8ubpRhGvifZ2YzwvTewD7ALWH+vBDLu9PMj2aXoNgyEM85rACON7OZ5h8vtQ9wVchrDnCPc+7PeRmAnPc4t18O/Av+R7A/RqsacR5T13+IaQ+a2aHh+/2NtN1UahOa2Z5mtlOYng0cBmwsKAM6/X/QzJZk5L1PNPs3wE1heZ1zoaPMKshzJUp95oqQ8ynbG/BFYINz7uNdqxtxHjMN/MdtkGNpj4tkjmUVfGd1lFkFea5EZtuguy5FyPlUrGCcNkzXdl6FIR+f2G+lYxfmk/V5+mUSPefQePu7DpPo3BL2efplDPzPNrOZYXou/jHNvwrzleM9Q4gtk+g4h4uBI0Kes/F9zotDXgY8mTAeOUgm0bmZHQh8Dh9X4ov8RsJ5HkXHw3qPW/8S2Ces3wE4Hn8csjyNzRjKoBhXzwU03p9vknF1HmJDsjGUQTFC/kdiDGVQjKvnApL356swzs5thMZQeuKcG5kXPhi/NJpfCTwtmt8M/A5/YexvgH3xF7FdDawD1gOfxF9UBf4K7W8Bm/Adzb3D8sXAxRnl7w2sDa/1wGnRujmhPjcBlwJPDMuPAz7XY39+hr+t9tZQ3yOjcq4K9foWMDNKswaYE6bPxF+U0J3vu0L9rgX+F/8oCfC39d6YsX23o/XAEdH6B4HF8tyf57Duw6EOj4X308Pyc4GfRo5+QTi3Q/l3ynmtc3tP/J1aN4Q01wInZzi6G3h9lO424E0656fkuzzUby2wCnheWP444FZgp4w0PwGuw39xbQKOjtbdAxwlz414fmuowyPAb4EvhOVfxHdQbg7H4SraseWd+Atn5bye80eC11ZseU9YbsCnI+fx9+c7gX/KyGvs2jdh3YXA/tG+nZqR9xtox+01wKsqxo2vALOyyqz6kufScaPyOSznpZ03Fjfkn9d0+T8mWnctsGdGmi8Hv+vwHe750boz4zzkuS/PS8J+PoTvw1wcrTsNf55vBF4RLT8O+Jic13a+Cd9/acWWs5p23q//yOf1oT5nApbnjB5tQuBl4TiuDe9xv61XGdvj+4TbRcfiuRl5fzLyfxmwXx/nQkeZ8lzfc1j3dfxj3P8c9vekKp85Oa98bh+GH0tZRzu2HNW082nk/4O0+5uXAc8Iy8uO0277zgrrt/Vx5bm+57Cu1zhtZtsgrKvUbpTzju3LjtP25XwcPhPxPtU8dkn7PPI8Jc3A+zxyPiVN0j7PhPt/Hj5OrA3vJ0XrKsX7sK72uJUcd6Q5ONTzD/jfbNZH604Mn4FNwN9HyxcD35bz2s4vxcfwVlxZMWrO6x4Peoxbh3VH4S+mupnOMYxtnhjDMRR5npJm4GMoct6xffIxlAn3PzJjKPI8Jc3Ax1DkvGP7kRtDyXQ77IPbJe0g4MvR/BLgAwMoZzn+n2lN5PUR4FkN5XVg1/5fCOxQIf3RwFureA5lXi7PzXvu5TzMLwE+gL8T7yoaakTLebH/4HwjcEYT+zZN/D8T+HhFz7vjG3eKLQ177uU8zG/73gSuAGbLebPOC/LLdJ53nJp8NXycdgdWRvNfAeYN0m13mTXqLM/916dS3JDz9M7ln12Bb1Uss6/2tjw3Up/vAE+X89F1PiT/TbYJlwDvj+Yzf3Rp8lzoLlOeB+M5J6/K7UY5T+9c/muNH3b0ceV5MJ5z8qrcbpTz9M4r5p/8+HTvU9VjxxD6PPLcSN1qt7/lfDycj7v/kKZSvA9pGvuNQY4r1+2TwEvkfHKdpzwedP5BZuzGUOS57/3sqz8v5+PtfIz8Jx9Dkee+93Oo/VA5T+R22Ac3Q8KJdP6z4ORh1ynhvr8MWJjSc6tMeU7nPJo/Gf+4phcO28mkOs/y33Kuc37gng8GDpDndM6j+ZOBeUT/sprU1yjElqguuc7HrX3T+gyPW5ny3FddasUNOU/vXP4rl9l3e1ue+6rLDsAb5Xz0nY+5/9cCu41bmfLcV11qfebkPL1z+a9cZt99XHnuqy612o1ynt75KB+fFPs0CmVOR885dem7/S3n4+F8uvlnyL8xTAfHBfX5BzmffOfDOB5D6ueMTPtzkj3n1GVkxi3lfPivSfLPCF0nMMmec+oyat/jcj6AV+sRFEIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBBCiGnK44ZdASGEEEIIIYQQQgghhBBCCCGEEEIIIYQQQgghxHDRRcVCCCGEEEIIIYQQQgghhBBCCCGEEEIIIYQQQkxzdFGxEEIIIYQQQgghhBBCCCGEEEIIIYQQQgghhBDTHF1ULIQQQgghhBBCCCGEEEIIIYQQQgghhBBCCCHENEcXFQshhBBCCCGEEEIIIYQQQgghhBBCCCGEEEIIMc35f4kIV0Uk6WSyAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 3600x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "prec_recall_scores = []\n",
    "for params in param_combos:\n",
    "    folds = all_scores[params]\n",
    "    prec_recall_scores.append([fold[0] for fold in folds])\n",
    "param_combos_str = [str(params) for params in param_combos]\n",
    "plt.figure(figsize=(50,10))\n",
    "sns.boxplot(param_combos_str, prec_recall_scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# best => kernel_width = 3, hidden_layers = (30,), # kernels = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[0.8530917439307213,\n",
       "  0.779962295929569,\n",
       "  0.9989452154053583,\n",
       "  0.8249638689110694],\n",
       " [0.8100564718249725,\n",
       "  0.805814123257335,\n",
       "  0.998385706161833,\n",
       "  0.7849615199361852],\n",
       " [0.8229270145064216,\n",
       "  0.7078166369661463,\n",
       "  0.9988259573492319,\n",
       "  0.7759298800793263]]"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_scores[((30,), 3, 10)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1992ac3c8>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 3600x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "aucs = []\n",
    "for params in param_combos:\n",
    "    folds = all_scores[params]\n",
    "    aucs.append([fold[1] for fold in folds])\n",
    "param_combos_str = [str(params) for params in param_combos]\n",
    "plt.figure(figsize=(50,10))\n",
    "sns.boxplot(param_combos_str, aucs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}