fx_.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "metadata": {
    "id": "SEuXjbiwtHh7",
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt"
   ],
   "execution_count": 1,
   "outputs": []
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "outputs": [],
   "source": [
    "# Чтение базы\n",
    "massiv_fx = pd.read_csv('massiv.csv')"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "outputs": [
    {
     "data": {
      "text/plain": "       Time     Open    Close      Low     High    RSI     MFI    MACD  \\\n0      2145  111.843  111.828  111.797  111.871  43.65  44.199  0.1425   \n1      2200  111.826  111.836  111.799  111.879  43.74  43.872  0.1443   \n2      2215  111.835  111.950  111.814  111.971  45.00  44.942  0.1443   \n3      2230  111.951  112.005  111.928  112.033  45.59  45.927  0.1428   \n4      2245  112.005  112.047  111.983  112.078  46.04  47.031  0.1405   \n...     ...      ...      ...      ...      ...    ...     ...     ...   \n99894  1245  118.115  118.100  118.088  118.167  44.43  49.984  0.0144   \n99895  1300  118.099  118.138  118.094  118.149  45.14  49.956  0.0205   \n99896  1315  118.137  118.134  118.112  118.191  45.08  50.000  0.0259   \n99897  1330  118.135  118.216  118.122  118.224  46.58  49.891  0.0300   \n99898  1345  118.216  118.208  118.202  118.269  46.45  51.144  0.0334   \n\n       Stochastik  \n0           13.57  \n1           13.60  \n2           13.73  \n3           13.94  \n4           14.18  \n...           ...  \n99894       67.87  \n99895       67.50  \n99896       67.06  \n99897       66.81  \n99898       66.52  \n\n[99899 rows x 9 columns]",
      "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>Time</th>\n      <th>Open</th>\n      <th>Close</th>\n      <th>Low</th>\n      <th>High</th>\n      <th>RSI</th>\n      <th>MFI</th>\n      <th>MACD</th>\n      <th>Stochastik</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>2145</td>\n      <td>111.843</td>\n      <td>111.828</td>\n      <td>111.797</td>\n      <td>111.871</td>\n      <td>43.65</td>\n      <td>44.199</td>\n      <td>0.1425</td>\n      <td>13.57</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2200</td>\n      <td>111.826</td>\n      <td>111.836</td>\n      <td>111.799</td>\n      <td>111.879</td>\n      <td>43.74</td>\n      <td>43.872</td>\n      <td>0.1443</td>\n      <td>13.60</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>2215</td>\n      <td>111.835</td>\n      <td>111.950</td>\n      <td>111.814</td>\n      <td>111.971</td>\n      <td>45.00</td>\n      <td>44.942</td>\n      <td>0.1443</td>\n      <td>13.73</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>2230</td>\n      <td>111.951</td>\n      <td>112.005</td>\n      <td>111.928</td>\n      <td>112.033</td>\n      <td>45.59</td>\n      <td>45.927</td>\n      <td>0.1428</td>\n      <td>13.94</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>2245</td>\n      <td>112.005</td>\n      <td>112.047</td>\n      <td>111.983</td>\n      <td>112.078</td>\n      <td>46.04</td>\n      <td>47.031</td>\n      <td>0.1405</td>\n      <td>14.18</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>99894</th>\n      <td>1245</td>\n      <td>118.115</td>\n      <td>118.100</td>\n      <td>118.088</td>\n      <td>118.167</td>\n      <td>44.43</td>\n      <td>49.984</td>\n      <td>0.0144</td>\n      <td>67.87</td>\n    </tr>\n    <tr>\n      <th>99895</th>\n      <td>1300</td>\n      <td>118.099</td>\n      <td>118.138</td>\n      <td>118.094</td>\n      <td>118.149</td>\n      <td>45.14</td>\n      <td>49.956</td>\n      <td>0.0205</td>\n      <td>67.50</td>\n    </tr>\n    <tr>\n      <th>99896</th>\n      <td>1315</td>\n      <td>118.137</td>\n      <td>118.134</td>\n      <td>118.112</td>\n      <td>118.191</td>\n      <td>45.08</td>\n      <td>50.000</td>\n      <td>0.0259</td>\n      <td>67.06</td>\n    </tr>\n    <tr>\n      <th>99897</th>\n      <td>1330</td>\n      <td>118.135</td>\n      <td>118.216</td>\n      <td>118.122</td>\n      <td>118.224</td>\n      <td>46.58</td>\n      <td>49.891</td>\n      <td>0.0300</td>\n      <td>66.81</td>\n    </tr>\n    <tr>\n      <th>99898</th>\n      <td>1345</td>\n      <td>118.216</td>\n      <td>118.208</td>\n      <td>118.202</td>\n      <td>118.269</td>\n      <td>46.45</td>\n      <td>51.144</td>\n      <td>0.0334</td>\n      <td>66.52</td>\n    </tr>\n  </tbody>\n</table>\n<p>99899 rows × 9 columns</p>\n</div>"
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "massiv_fx"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "outputs": [],
   "source": [
    "massiv_fx_len = massiv_fx.shape[0] // 9\n",
    "massiv_blok = 100 # Формируем матрицы для анализа 96*4*24"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "# Делим на базы\n",
    "bd_train = massiv_fx[:massiv_fx_len * 6].reset_index(drop=True)\n",
    "bd_test = massiv_fx[massiv_fx_len * 6 + massiv_blok * 2:massiv_fx_len * 8].reset_index(drop=True)\n",
    "bd_val = massiv_fx[massiv_fx_len * 8 + massiv_blok * 2:].reset_index(drop=True)"
   ],
   "metadata": {
    "id": "RP5orUBov-Qs"
   },
   "execution_count": 11,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "bd_train\n",
    "bd_test\n",
    "bd_val"
   ],
   "metadata": {
    "id": "Re_XGxIBBC9c"
   },
   "execution_count": 12,
   "outputs": [
    {
     "data": {
      "text/plain": "       Time     Open    Close      Low     High    RSI     MFI    MACD  \\\n0      1315  113.979  113.932  113.901  113.986  43.42  45.219  0.1042   \n1      1330  113.929  113.905  113.899  113.951  43.08  45.304  0.1053   \n2      1345  113.906  114.003  113.900  114.040  44.66  46.616  0.1050   \n3      1400  114.001  113.974  113.960  114.046  44.29  47.985  0.1045   \n4      1415  113.973  113.979  113.944  114.021  44.37  47.921  0.1042   \n...     ...      ...      ...      ...      ...    ...     ...     ...   \n10902  1245  118.115  118.100  118.088  118.167  44.43  49.984  0.0144   \n10903  1300  118.099  118.138  118.094  118.149  45.14  49.956  0.0205   \n10904  1315  118.137  118.134  118.112  118.191  45.08  50.000  0.0259   \n10905  1330  118.135  118.216  118.122  118.224  46.58  49.891  0.0300   \n10906  1345  118.216  118.208  118.202  118.269  46.45  51.144  0.0334   \n\n       Stochastik  \n0           16.14  \n1           16.19  \n2           16.34  \n3           16.58  \n4           16.86  \n...           ...  \n10902       67.87  \n10903       67.50  \n10904       67.06  \n10905       66.81  \n10906       66.52  \n\n[10907 rows x 9 columns]",
      "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>Time</th>\n      <th>Open</th>\n      <th>Close</th>\n      <th>Low</th>\n      <th>High</th>\n      <th>RSI</th>\n      <th>MFI</th>\n      <th>MACD</th>\n      <th>Stochastik</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1315</td>\n      <td>113.979</td>\n      <td>113.932</td>\n      <td>113.901</td>\n      <td>113.986</td>\n      <td>43.42</td>\n      <td>45.219</td>\n      <td>0.1042</td>\n      <td>16.14</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>1330</td>\n      <td>113.929</td>\n      <td>113.905</td>\n      <td>113.899</td>\n      <td>113.951</td>\n      <td>43.08</td>\n      <td>45.304</td>\n      <td>0.1053</td>\n      <td>16.19</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>1345</td>\n      <td>113.906</td>\n      <td>114.003</td>\n      <td>113.900</td>\n      <td>114.040</td>\n      <td>44.66</td>\n      <td>46.616</td>\n      <td>0.1050</td>\n      <td>16.34</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>1400</td>\n      <td>114.001</td>\n      <td>113.974</td>\n      <td>113.960</td>\n      <td>114.046</td>\n      <td>44.29</td>\n      <td>47.985</td>\n      <td>0.1045</td>\n      <td>16.58</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>1415</td>\n      <td>113.973</td>\n      <td>113.979</td>\n      <td>113.944</td>\n      <td>114.021</td>\n      <td>44.37</td>\n      <td>47.921</td>\n      <td>0.1042</td>\n      <td>16.86</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>10902</th>\n      <td>1245</td>\n      <td>118.115</td>\n      <td>118.100</td>\n      <td>118.088</td>\n      <td>118.167</td>\n      <td>44.43</td>\n      <td>49.984</td>\n      <td>0.0144</td>\n      <td>67.87</td>\n    </tr>\n    <tr>\n      <th>10903</th>\n      <td>1300</td>\n      <td>118.099</td>\n      <td>118.138</td>\n      <td>118.094</td>\n      <td>118.149</td>\n      <td>45.14</td>\n      <td>49.956</td>\n      <td>0.0205</td>\n      <td>67.50</td>\n    </tr>\n    <tr>\n      <th>10904</th>\n      <td>1315</td>\n      <td>118.137</td>\n      <td>118.134</td>\n      <td>118.112</td>\n      <td>118.191</td>\n      <td>45.08</td>\n      <td>50.000</td>\n      <td>0.0259</td>\n      <td>67.06</td>\n    </tr>\n    <tr>\n      <th>10905</th>\n      <td>1330</td>\n      <td>118.135</td>\n      <td>118.216</td>\n      <td>118.122</td>\n      <td>118.224</td>\n      <td>46.58</td>\n      <td>49.891</td>\n      <td>0.0300</td>\n      <td>66.81</td>\n    </tr>\n    <tr>\n      <th>10906</th>\n      <td>1345</td>\n      <td>118.216</td>\n      <td>118.208</td>\n      <td>118.202</td>\n      <td>118.269</td>\n      <td>46.45</td>\n      <td>51.144</td>\n      <td>0.0334</td>\n      <td>66.52</td>\n    </tr>\n  </tbody>\n</table>\n<p>10907 rows × 9 columns</p>\n</div>"
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ]
  },
  {
   "cell_type": "markdown",
   "source": [
    "Стандартизация"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler, RobustScaler"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "outputs": [],
   "source": [
    "ss = RobustScaler()\n",
    "bd_train_sc = ss.fit_transform(bd_train)\n",
    "bd_test_sc = ss.fit_transform(bd_test)\n",
    "bd_val_sc = ss.fit_transform(bd_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "outputs": [
    {
     "data": {
      "text/plain": "array([[ 7.87500000e-01,  9.09950366e-03,  5.55424250e-03,\n         6.87040986e-03,  7.54716981e-03, -1.09773124e+00,\n        -7.17366496e-01,  1.32242991e+00, -1.00408497e+00],\n       [ 8.33333333e-01,  5.08154101e-03,  7.44504845e-03,\n         7.34423122e-03,  9.43396226e-03, -1.08202443e+00,\n        -7.55640088e-01,  1.33925234e+00, -1.00326797e+00],\n       [ 8.45833333e-01,  7.20869771e-03,  3.43890333e-02,\n         1.08978915e-02,  3.11320755e-02, -8.62129145e-01,\n        -6.30402341e-01,  1.33925234e+00, -9.99727669e-01],\n       [ 8.58333333e-01,  3.46253841e-02,  4.73883243e-02,\n         3.79057095e-02,  4.57547170e-02, -7.59162304e-01,\n        -5.15113387e-01,  1.32523364e+00, -9.94008715e-01],\n       [ 8.70833333e-01,  4.73883243e-02,  5.73150555e-02,\n         5.09357972e-02,  5.63679245e-02, -6.80628272e-01,\n        -3.85896123e-01,  1.30373832e+00, -9.87472767e-01],\n       [ 9.16666667e-01,  5.25880407e-02,  6.93689435e-02,\n         6.04122246e-02,  6.39150943e-02, -5.88132635e-01,\n        -2.70958303e-01,  1.27476636e+00, -9.79575163e-01],\n       [ 9.29166667e-01,  6.96052943e-02,  6.25147719e-02,\n         6.98886520e-02,  6.29716981e-02, -6.33507853e-01,\n        -3.16605706e-01,  1.24766355e+00, -9.71677560e-01],\n       [ 9.41666667e-01,  6.18057197e-02,  6.39328764e-02,\n         6.91779199e-02,  5.66037736e-02, -6.23036649e-01,\n        -2.65340161e-01,  1.22149533e+00, -9.63507625e-01],\n       [ 9.54166667e-01,  6.39328764e-02,  6.84235405e-02,\n         7.27315802e-02,  6.32075472e-02, -5.86387435e-01,\n        -1.62692026e-01,  1.19252336e+00, -9.57516340e-01],\n       [-1.00000000e+00,  7.78775703e-02,  7.12597495e-02,\n         7.81805259e-02,  7.38207547e-02, -5.65445026e-01,\n        -2.04594001e-01,  1.15981308e+00, -9.50708061e-01],\n       [-9.87500000e-01,  7.07870480e-02,  8.02410778e-02,\n         7.93650794e-02,  7.19339623e-02, -4.93891798e-01,\n        -2.14074616e-01,  1.12523364e+00, -9.45806100e-01],\n       [-9.75000000e-01,  8.02410778e-02,  7.57504136e-02,\n         7.81805259e-02,  8.65566038e-02, -5.25305410e-01,\n        -1.83174835e-01,  1.09158879e+00, -9.40904139e-01],\n       [-9.62500000e-01,  7.69321673e-02,  8.75679508e-02,\n         8.29187396e-02,  8.13679245e-02, -4.31064572e-01,\n        -1.16108266e-01,  1.05420561e+00, -9.33551198e-01],\n       [-9.16666667e-01,  8.87497046e-02,  9.01678090e-02,\n         8.31556503e-02,  8.91509434e-02, -4.10122164e-01,\n        -3.80395026e-02,  1.01588785e+00, -9.24291939e-01],\n       [-9.04166667e-01,  8.89860553e-02,  8.66225479e-02,\n         9.42904525e-02,  9.55188679e-02, -4.36300175e-01,\n         2.30577908e-02,  9.76635514e-01, -9.15849673e-01],\n       [-8.91666667e-01,  8.61498464e-02,  8.00047270e-02,\n         8.67093106e-02,  8.06603774e-02, -4.83420593e-01,\n        -6.62472568e-02,  9.45794393e-01, -9.09313725e-01],\n       [-8.79166667e-01,  7.97683763e-02,  9.46584732e-02,\n         8.81307747e-02,  8.77358491e-02, -3.66492147e-01,\n        -7.57278713e-02,  9.09345794e-01, -9.01416122e-01],\n       [-8.33333333e-01,  9.41857717e-02,  7.71685181e-02,\n         8.50509358e-02,  8.67924528e-02, -4.92146597e-01,\n        -6.20336503e-02,  8.81308411e-01, -8.94335512e-01],\n       [-8.20833333e-01,  7.71685181e-02,  9.58402269e-02,\n         8.41032931e-02,  9.17452830e-02, -3.43804538e-01,\n         1.75566935e-03,  8.45794393e-01, -8.85893246e-01],\n       [-8.08333333e-01,  9.60765776e-02,  1.03639801e-01,\n         1.00687041e-01,  9.74056604e-02, -2.82722513e-01,\n         9.62106803e-02,  8.05607477e-01, -8.72004357e-01],\n       [-7.95833333e-01,  1.03876152e-01,  9.56038762e-02,\n         1.04477612e-01,  1.03537736e-01, -3.42059337e-01,\n         2.27183614e-01,  7.68224299e-01, -8.61383442e-01],\n       [-7.50000000e-01,  9.58402269e-02,  8.68588986e-02,\n         9.28689884e-02,  8.96226415e-02, -4.04886562e-01,\n         9.25822970e-02,  7.40186916e-01, -8.52124183e-01],\n       [-7.37500000e-01,  8.70952494e-02,  8.54407941e-02,\n         8.24449183e-02,  7.92452830e-02, -4.15357766e-01,\n        -1.79078274e-02,  7.16822430e-01, -8.44226580e-01],\n       [-7.25000000e-01,  8.54407941e-02,  1.05530607e-01,\n         8.86045961e-02,  1.01415094e-01, -2.56544503e-01,\n         1.23365033e-01,  6.82242991e-01, -8.32516340e-01],\n       [-7.12500000e-01,  1.05057906e-01,  1.19948003e-01,\n         1.09689647e-01,  1.23113208e-01, -1.43106457e-01,\n         2.60658376e-01,  6.36448598e-01, -8.19172113e-01],\n       [-6.66666667e-01,  1.19948003e-01,  1.02930749e-01,\n         1.08505094e-01,  1.14622642e-01, -2.68760908e-01,\n         1.17863936e-01,  6.00000000e-01, -8.09640523e-01],\n       [-6.54166667e-01,  1.03167100e-01,  1.06239660e-01,\n         1.05188344e-01,  1.01179245e-01, -2.42582897e-01,\n        -2.01316752e-02,  5.67289720e-01, -7.98747277e-01],\n       [-6.41666667e-01,  1.05766958e-01,  1.30583786e-01,\n         1.11111111e-01,  1.29716981e-01, -5.58464223e-02,\n         2.34089247e-04,  5.20560748e-01, -7.86492375e-01],\n       [-6.29166667e-01,  1.30347436e-01,  1.13330182e-01,\n         1.20824449e-01,  1.24056604e-01, -1.83246073e-01,\n         2.17702999e-02,  4.82242991e-01, -7.73692810e-01],\n       [-5.83333333e-01,  1.13802883e-01,  1.11439376e-01,\n         1.18929164e-01,  1.13207547e-01, -1.97207679e-01,\n         4.89246525e-02,  4.48598131e-01, -7.61710240e-01],\n       [-5.70833333e-01,  1.11439376e-01,  1.17111794e-01,\n         1.15612414e-01,  1.08962264e-01, -1.53577661e-01,\n        -5.26700805e-02,  4.15887850e-01, -7.47821351e-01],\n       [-5.58333333e-01,  1.16875443e-01,  1.17111794e-01,\n         1.15138593e-01,  1.10849057e-01, -1.53577661e-01,\n         5.99268471e-02,  3.84112150e-01, -7.33115468e-01],\n       [-5.45833333e-01,  1.17111794e-01,  1.07421413e-01,\n         1.11821843e-01,  1.08962264e-01, -2.25130890e-01,\n        -3.28895391e-02,  3.58878505e-01, -7.20043573e-01],\n       [-5.00000000e-01,  1.07657764e-01,  9.58402269e-02,\n         1.04714523e-01,  1.00000000e-01, -3.12390925e-01,\n         4.09656181e-03,  3.41121495e-01, -7.05610022e-01],\n       [-4.87500000e-01,  9.58402269e-02,  8.94587568e-02,\n         9.66595593e-02,  9.19811321e-02, -3.59511344e-01,\n         3.23043160e-02,  3.29906542e-01, -6.92265795e-01],\n       [-4.75000000e-01,  8.96951075e-02,  9.06405105e-02,\n         9.52380952e-02,  8.41981132e-02, -3.50785340e-01,\n        -4.19019751e-02,  3.20560748e-01, -6.80555556e-01],\n       [-4.62500000e-01,  9.08768613e-02,  9.08768613e-02,\n         9.28689884e-02,  8.37264151e-02, -3.47294939e-01,\n        -1.16576445e-01,  3.12149533e-01, -6.68845316e-01],\n       [-4.16666667e-01,  9.11132120e-02,  8.80406523e-02,\n         9.68964700e-02,  9.12735849e-02, -3.69982548e-01,\n        -3.61667886e-02,  3.02803738e-01, -6.57407407e-01],\n       [-4.04166667e-01,  8.75679508e-02,  8.61498464e-02,\n         9.35797204e-02,  8.16037736e-02, -3.83944154e-01,\n        -1.08617410e-01,  2.96261682e-01, -6.47058824e-01],\n       [-3.91666667e-01,  8.56771449e-02,  8.49680926e-02,\n         9.07367922e-02,  8.01886792e-02, -3.92670157e-01,\n        -1.92187271e-01,  2.91588785e-01, -6.37527233e-01],\n       [-3.79166667e-01,  8.52044434e-02,  8.16591822e-02,\n         8.83676854e-02,  8.16037736e-02, -4.18848168e-01,\n        -1.87154353e-01,  2.88785047e-01, -6.27178649e-01],\n       [-3.33333333e-01,  8.16591822e-02,  8.56771449e-02,\n         8.36294717e-02,  7.83018868e-02, -3.83944154e-01,\n        -2.60892465e-01,  2.85046729e-01, -6.16285403e-01],\n       [-3.20833333e-01,  8.54407941e-02,  8.49680926e-02,\n         8.74200426e-02,  7.87735849e-02, -3.89179756e-01,\n        -1.90782736e-01,  2.81308411e-01, -6.05936819e-01],\n       [-3.08333333e-01,  8.47317419e-02,  9.86764358e-02,\n         9.35797204e-02,  9.12735849e-02, -2.72251309e-01,\n        -1.98741770e-01,  2.69158879e-01, -5.95043573e-01],\n       [-2.95833333e-01,  9.84400851e-02,  1.01039943e-01,\n         9.90286662e-02,  9.43396226e-02, -2.51308901e-01,\n        -1.16810534e-01,  2.54205607e-01, -5.82516340e-01],\n       [-2.50000000e-01,  1.00803593e-01,  8.82770031e-02,\n         9.66595593e-02,  9.24528302e-02, -3.54275742e-01,\n        -1.07798098e-01,  2.45794393e-01, -5.71078431e-01],\n       [-2.37500000e-01,  8.82770031e-02,  9.74946821e-02,\n         9.54750059e-02,  9.03301887e-02, -2.75741710e-01,\n        -1.23716167e-01,  2.35514019e-01, -5.59095861e-01],\n       [-2.25000000e-01,  9.74946821e-02,  8.87497046e-02,\n         9.54750059e-02,  8.98584906e-02, -3.45549738e-01,\n        -1.21258230e-01,  2.28037383e-01, -5.47385621e-01],\n       [-2.12500000e-01,  8.87497046e-02,  9.37130702e-02,\n         9.64226487e-02,  8.86792453e-02, -3.01919721e-01,\n        -2.37600585e-02,  2.19626168e-01, -5.32135076e-01],\n       [-1.66666667e-01,  9.34767195e-02,  7.36232569e-02,\n         7.67590618e-02,  8.51415094e-02, -4.62478185e-01,\n        -1.19385516e-02,  2.23364486e-01, -5.19880174e-01],\n       [-1.54166667e-01,  7.38596077e-02,  4.66792720e-02,\n         5.35418147e-02,  6.76886792e-02, -6.68411867e-01,\n        -1.15288954e-01,  2.43925234e-01, -5.10348584e-01],\n       [-1.41666667e-01,  4.71519735e-02,  4.33703616e-02,\n         4.14593698e-02,  3.96226415e-02, -6.92844677e-01,\n        -8.32187271e-02,  2.71962617e-01, -4.99727669e-01],\n       [-1.29166667e-01,  4.33703616e-02,  4.45521153e-02,\n         4.14593698e-02,  3.79716981e-02, -6.82373473e-01,\n        -2.21565472e-01,  2.97196262e-01, -4.91830065e-01],\n       [-8.33333333e-02,  4.38430631e-02,  4.00614512e-02,\n         3.57735134e-02,  4.00943396e-02, -7.17277487e-01,\n        -3.59561083e-01,  3.24299065e-01, -4.85566449e-01],\n       [-7.08333333e-02,  3.98251004e-02,  2.09170409e-02,\n         2.03743189e-02,  3.77358491e-02, -8.62129145e-01,\n        -5.09027067e-01,  3.58878505e-01, -4.83387800e-01],\n       [-5.83333333e-02,  2.11533916e-02,  1.77263058e-03,\n         7.81805259e-03,  2.28773585e-02, -1.00174520e+00,\n        -6.42457937e-01,  4.03738318e-01, -4.83387800e-01],\n       [-4.58333333e-02,  2.24533207e-03, -1.71354290e-02,\n        -1.70575693e-02,  3.30188679e-03, -1.13787086e+00,\n        -6.37307974e-01,  4.60747664e-01, -4.85566449e-01],\n       [ 0.00000000e+00, -1.71354290e-02, -1.42992200e-02,\n        -1.20824449e-02, -5.42452830e-03, -1.11169284e+00,\n        -6.04769568e-01,  5.14018692e-01, -4.85021786e-01],\n       [ 1.25000000e-02, -1.42992200e-02,  1.06357835e-03,\n        -1.23193556e-02, -2.59433962e-03, -9.70331588e-01,\n        -4.49100219e-01,  5.57009346e-01, -4.79847495e-01],\n       [ 2.50000000e-02,  1.53627984e-03, -1.40628693e-02,\n        -5.44894575e-03,  6.36792453e-03, -1.07853403e+00,\n        -3.17073884e-01,  6.00934579e-01, -4.75217865e-01],\n       [ 3.75000000e-02, -1.42992200e-02, -4.02978019e-02,\n        -3.43520493e-02, -1.46226415e-02, -1.26178010e+00,\n        -2.97059254e-01,  6.59813084e-01, -4.70588235e-01],\n       [ 8.33333333e-02, -4.00614512e-02, -5.90876861e-04,\n        -3.15091211e-02, -8.49056604e-03, -9.09249564e-01,\n        -1.43964887e-01,  6.98130841e-01, -4.59422658e-01],\n       [ 9.58333333e-02, -5.90876861e-04, -1.92625857e-02,\n        -2.34541578e-02, -7.78301887e-03, -1.03839442e+00,\n        -1.31324067e-01,  7.40186916e-01, -4.50980392e-01],\n       [ 1.08333333e-01, -1.87898842e-02,  6.49964547e-03,\n        -2.86661928e-02,  6.83962264e-03, -8.18499127e-01,\n        -1.13065106e-01,  7.67289720e-01, -4.42265795e-01],\n       [ 1.20833333e-01,  6.73599622e-03,  1.38265186e-02,\n         6.63349917e-03,  1.29716981e-02, -7.55671902e-01,\n        -1.19034382e-01,  7.82242991e-01, -4.34912854e-01],\n       [ 1.66666667e-01,  1.35901678e-02,  1.14630111e-02,\n         1.56361052e-02,  1.39150943e-02, -7.73123909e-01,\n         1.15874177e-02,  7.94392523e-01, -4.26198257e-01],\n       [ 1.79166667e-01,  1.14630111e-02,  1.05176081e-02,\n         9.00260602e-03,  6.60377358e-03, -7.80104712e-01,\n         2.18873446e-02,  8.08411215e-01, -4.17483660e-01],\n       [ 1.91666667e-01,  1.02812574e-02,  3.01347199e-02,\n         1.35039090e-02,  2.35849057e-02, -6.14310646e-01,\n         1.21258230e-01,  8.09345794e-01, -4.04956427e-01],\n       [ 2.04166667e-01,  3.03710707e-02,  7.20869771e-03,\n         1.58730159e-02,  2.31132075e-02, -7.76614311e-01,\n         2.01316752e-02,  8.17757009e-01, -3.96786492e-01],\n       [ 2.50000000e-01,  7.44504845e-03, -4.37248877e-03,\n         3.07983890e-03,  4.95283019e-03, -8.56893543e-01,\n        -9.19970739e-02,  8.35514019e-01, -3.91067538e-01],\n       [ 2.62500000e-01, -4.60883952e-03, -4.52611676e-02,\n        -4.09855484e-02, -1.25000000e-02, -1.13263525e+00,\n        -2.14425750e-01,  8.78504673e-01, -3.91339869e-01],\n       [ 2.75000000e-01, -4.83337273e-02, -3.55707870e-02,\n        -3.98009950e-02, -3.51415094e-02, -1.05061082e+00,\n        -2.45559620e-01,  9.17757009e-01, -3.89978214e-01],\n       [ 2.87500000e-01, -3.62798393e-02, -2.72985110e-02,\n        -4.43022980e-02, -2.09905660e-02, -9.80802792e-01,\n        -1.11192392e-01,  9.49532710e-01, -3.85620915e-01],\n       [ 3.33333333e-01, -2.70621602e-02, -6.98416450e-02,\n        -6.18336887e-02, -3.51415094e-02, -1.25654450e+00,\n        -1.43964887e-01,  1.00000000e+00, -3.86710240e-01],\n       [ 3.45833333e-01, -6.96052943e-02, -6.13330182e-02,\n        -6.70457238e-02, -6.62735849e-02, -1.18499127e+00,\n        -2.76459400e-01,  1.05046729e+00, -3.88616558e-01],\n       [ 3.58333333e-01, -6.10966675e-02, -5.02245332e-02,\n        -6.37289742e-02, -1.43867925e-02, -1.09249564e+00,\n        -2.29641551e-01,  1.08317757e+00, -3.88071895e-01],\n       [ 3.70833333e-01, -5.09335854e-02, -3.29709289e-02,\n        -4.35915660e-02, -1.79245283e-02, -9.51134380e-01,\n        -7.54937820e-02,  1.10373832e+00, -3.80718954e-01],\n       [ 4.16666667e-01, -3.29709289e-02,  1.14630111e-02,\n        -2.44018005e-02,  6.60377358e-03, -6.00349040e-01,\n        -5.85223116e-02,  1.09813084e+00, -3.69825708e-01],\n       [ 4.29166667e-01,  1.09903096e-02,  2.72985110e-02,\n        -1.18455342e-03,  2.02830189e-02, -4.79930192e-01,\n         1.04052670e-01,  1.07943925e+00, -3.55664488e-01],\n       [ 4.41666667e-01,  2.70621602e-02,  4.59702198e-02,\n         1.82421227e-02,  4.15094340e-02, -3.42059337e-01,\n         2.84301390e-01,  1.04579439e+00, -3.37690632e-01],\n       [ 4.54166667e-01,  4.57338691e-02,  2.91893169e-02,\n         3.76687989e-02,  4.90566038e-02, -4.55497382e-01,\n         4.59634236e-01,  1.01495327e+00, -3.19716776e-01],\n       [ 5.00000000e-01,  2.91893169e-02,  4.14795557e-02,\n         2.34541578e-02,  3.65566038e-02, -3.64746946e-01,\n         2.80555962e-01,  9.85046729e-01, -3.02559913e-01],\n       [ 5.12500000e-01,  4.24249586e-02,  6.65327346e-02,\n         4.47761194e-02,  5.82547170e-02, -1.83246073e-01,\n         2.78800293e-01,  9.39252336e-01, -2.84041394e-01],\n       [ 5.25000000e-01,  6.65327346e-02,  3.72252423e-02,\n         4.40653873e-02,  5.82547170e-02, -3.78708551e-01,\n         1.07798098e-01,  9.06542056e-01, -2.67701525e-01],\n       [ 5.37500000e-01,  3.76979437e-02,  3.39163318e-02,\n         3.60104241e-02,  4.69339623e-02, -4.01396161e-01,\n        -5.67666423e-02,  8.79439252e-01, -2.59531590e-01],\n       [ 5.83333333e-01,  3.36799811e-02,  4.76246750e-02,\n         2.65339967e-02,  4.38679245e-02, -3.01919721e-01,\n        -5.51280176e-02,  8.51401869e-01, -2.50816993e-01],\n       [ 5.95833333e-01,  4.73883243e-02,  4.80973765e-02,\n         4.57237621e-02,  5.87264151e-02, -2.98429319e-01,\n        -4.92757864e-02,  8.16822430e-01, -2.48910675e-01],\n       [ 6.08333333e-01,  4.80973765e-02,  4.66792720e-02,\n         4.50130301e-02,  4.19811321e-02, -3.07155323e-01,\n        -2.01199707e-01,  7.87850467e-01, -2.46187364e-01],\n       [ 6.20833333e-01,  4.66792720e-02,  3.36799811e-02,\n         3.52996920e-02,  4.05660377e-02, -3.96160558e-01,\n        -1.95230432e-01,  7.66355140e-01, -2.39379085e-01],\n       [ 6.66666667e-01,  3.39163318e-02, -9.80855590e-03,\n        -2.34541578e-02,  3.01886792e-02, -6.80628272e-01,\n        -2.40409656e-01,  7.74766355e-01, -2.37200436e-01],\n       [ 6.79166667e-01, -1.00449066e-02, -2.23351454e-02,\n        -1.77683014e-02, -7.07547170e-03, -7.59162304e-01,\n        -2.92845647e-01,  7.91588785e-01, -2.36111111e-01],\n       [ 6.91666667e-01, -2.32805483e-02, -5.08154101e-03,\n        -1.94266761e-02, -1.32075472e-02, -6.30017452e-01,\n        -1.30972933e-01,  8.02803738e-01, -2.32298475e-01],\n       [ 7.04166667e-01, -4.84519026e-03,  2.95438431e-03,\n        -1.18455342e-03,  8.72641509e-03, -5.72425829e-01,\n        -1.22896854e-01,  8.02803738e-01, -2.29847495e-01],\n       [ 7.50000000e-01,  3.19073505e-03,  6.26329473e-03,\n         6.63349917e-03,  5.18867925e-03, -5.47993019e-01,\n         2.53986832e-02,  7.99065421e-01, -2.21949891e-01],\n       [ 7.62500000e-01,  6.26329473e-03,  1.83171827e-02,\n         1.44515518e-02,  1.74528302e-02, -4.58987784e-01,\n         2.34089247e-03,  7.87850467e-01, -2.13507625e-01],\n       [ 7.75000000e-01,  1.83171827e-02,  2.02079887e-02,\n         2.29803364e-02,  2.02830189e-02, -4.45026178e-01,\n         1.43145574e-01,  7.72897196e-01, -2.00163399e-01],\n       [ 7.87500000e-01,  2.04443394e-02,  5.90876861e-04,\n         5.21203506e-03,  1.25000000e-02, -5.74171030e-01,\n         1.59531822e-01,  7.71028037e-01, -1.87908497e-01],\n       [ 8.33333333e-01,  3.54526117e-04, -1.28811156e-02,\n        -6.63349917e-03, -3.53773585e-03, -6.61431065e-01,\n         1.73577176e-01,  7.77570093e-01, -1.80827887e-01],\n       [ 8.45833333e-01, -1.31174663e-02, -2.65894588e-02,\n        -2.44018005e-02, -1.36792453e-02, -7.48691099e-01,\n         1.44316020e-01,  7.94392523e-01, -1.75653595e-01],\n       [ 8.58333333e-01, -2.68258095e-02, -5.54242496e-02,\n        -6.34920635e-02, -3.27830189e-02, -9.28446771e-01,\n        -1.76737381e-02,  8.30841121e-01, -1.76742919e-01]])"
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bd_train_sc[:100]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "outputs": [],
   "source": [
    "# Создаем матрицы для обучения\n",
    "X_train = []\n",
    "Y_train = []\n",
    "for col in range(len(bd_train_sc)-massiv_blok-1):\n",
    "    X_train.append(bd_train_sc[col:col+massiv_blok])\n",
    "    Y_train.append(bd_train_sc[col+massiv_blok,2])\n",
    "X_train = np.array(X_train)\n",
    "Y_train = np.array(Y_train)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "outputs": [],
   "source": [
    "# Создаем матрицы для тестирования\n",
    "X_test = []\n",
    "Y_test = []\n",
    "for col in range(len(bd_test_sc)-massiv_blok-1):\n",
    "    X_test.append(bd_test_sc[col:col+massiv_blok])\n",
    "    Y_test.append(bd_test_sc[col+massiv_blok,2])\n",
    "X_test = np.array(X_test)\n",
    "Y_test = np.array(Y_test)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "outputs": [],
   "source": [
    "# Создаем матрицы для валидации\n",
    "X_val = []\n",
    "Y_val = []\n",
    "for col in range(len(bd_val_sc)-massiv_blok-1):\n",
    "    X_val.append(bd_val_sc[col:col+massiv_blok])\n",
    "    Y_val.append(bd_val_sc[col+massiv_blok,2])\n",
    "X_val = np.array(X_val)\n",
    "Y_val = np.array(Y_val)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "outputs": [
    {
     "data": {
      "text/plain": "       Time     Open    Close      Low     High    RSI     MFI    MACD  \\\n0      2145  111.843  111.828  111.797  111.871  43.65  44.199  0.1425   \n1      2200  111.826  111.836  111.799  111.879  43.74  43.872  0.1443   \n2      2215  111.835  111.950  111.814  111.971  45.00  44.942  0.1443   \n3      2230  111.951  112.005  111.928  112.033  45.59  45.927  0.1428   \n4      2245  112.005  112.047  111.983  112.078  46.04  47.031  0.1405   \n...     ...      ...      ...      ...      ...    ...     ...     ...   \n66589  1930  109.928  109.958  109.908  109.974  57.69  46.724 -0.0767   \n66590  1945  109.957  109.953  109.932  109.990  57.56  46.419 -0.0787   \n66591  2000  109.952  109.955  109.940  109.980  57.60  47.827 -0.0806   \n66592  2015  109.955  109.971  109.937  109.983  57.91  47.411 -0.0825   \n66593  2030  109.972  109.888  109.877  109.979  55.75  47.230 -0.0831   \n\n       Stochastik  \n0           13.57  \n1           13.60  \n2           13.73  \n3           13.94  \n4           14.18  \n...           ...  \n66589       68.34  \n66590       68.92  \n66591       69.44  \n66592       70.07  \n66593       70.45  \n\n[66594 rows x 9 columns]",
      "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>Time</th>\n      <th>Open</th>\n      <th>Close</th>\n      <th>Low</th>\n      <th>High</th>\n      <th>RSI</th>\n      <th>MFI</th>\n      <th>MACD</th>\n      <th>Stochastik</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>2145</td>\n      <td>111.843</td>\n      <td>111.828</td>\n      <td>111.797</td>\n      <td>111.871</td>\n      <td>43.65</td>\n      <td>44.199</td>\n      <td>0.1425</td>\n      <td>13.57</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2200</td>\n      <td>111.826</td>\n      <td>111.836</td>\n      <td>111.799</td>\n      <td>111.879</td>\n      <td>43.74</td>\n      <td>43.872</td>\n      <td>0.1443</td>\n      <td>13.60</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>2215</td>\n      <td>111.835</td>\n      <td>111.950</td>\n      <td>111.814</td>\n      <td>111.971</td>\n      <td>45.00</td>\n      <td>44.942</td>\n      <td>0.1443</td>\n      <td>13.73</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>2230</td>\n      <td>111.951</td>\n      <td>112.005</td>\n      <td>111.928</td>\n      <td>112.033</td>\n      <td>45.59</td>\n      <td>45.927</td>\n      <td>0.1428</td>\n      <td>13.94</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>2245</td>\n      <td>112.005</td>\n      <td>112.047</td>\n      <td>111.983</td>\n      <td>112.078</td>\n      <td>46.04</td>\n      <td>47.031</td>\n      <td>0.1405</td>\n      <td>14.18</td>\n    </tr>\n    <tr>\n      <th>...</th>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n      <td>...</td>\n    </tr>\n    <tr>\n      <th>66589</th>\n      <td>1930</td>\n      <td>109.928</td>\n      <td>109.958</td>\n      <td>109.908</td>\n      <td>109.974</td>\n      <td>57.69</td>\n      <td>46.724</td>\n      <td>-0.0767</td>\n      <td>68.34</td>\n    </tr>\n    <tr>\n      <th>66590</th>\n      <td>1945</td>\n      <td>109.957</td>\n      <td>109.953</td>\n      <td>109.932</td>\n      <td>109.990</td>\n      <td>57.56</td>\n      <td>46.419</td>\n      <td>-0.0787</td>\n      <td>68.92</td>\n    </tr>\n    <tr>\n      <th>66591</th>\n      <td>2000</td>\n      <td>109.952</td>\n      <td>109.955</td>\n      <td>109.940</td>\n      <td>109.980</td>\n      <td>57.60</td>\n      <td>47.827</td>\n      <td>-0.0806</td>\n      <td>69.44</td>\n    </tr>\n    <tr>\n      <th>66592</th>\n      <td>2015</td>\n      <td>109.955</td>\n      <td>109.971</td>\n      <td>109.937</td>\n      <td>109.983</td>\n      <td>57.91</td>\n      <td>47.411</td>\n      <td>-0.0825</td>\n      <td>70.07</td>\n    </tr>\n    <tr>\n      <th>66593</th>\n      <td>2030</td>\n      <td>109.972</td>\n      <td>109.888</td>\n      <td>109.877</td>\n      <td>109.979</td>\n      <td>55.75</td>\n      <td>47.230</td>\n      <td>-0.0831</td>\n      <td>70.45</td>\n    </tr>\n  </tbody>\n</table>\n<p>66594 rows × 9 columns</p>\n</div>"
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bd_train"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "outputs": [
    {
     "data": {
      "text/plain": "array([[[ 0.09583333, -3.0813094 , -3.12823219, ..., -0.61696982,\n          1.1421371 , -1.04075574],\n        [ 0.10833333, -3.13410771, -3.15672823, ..., -0.6080328 ,\n          1.15322581, -1.03940621],\n        [ 0.12083333, -3.15839493, -3.05329815, ..., -0.47008727,\n          1.15020161, -1.03535762],\n        ...,\n        [ 0.12083333, -2.98627244, -2.93509235, ...,  0.28724635,\n          0.        , -0.07746289],\n        [ 0.16666667, -2.93664203, -2.92453826, ...,  0.2736831 ,\n          0.00907258, -0.07287449],\n        [ 0.17916667, -2.9281943 , -2.95831135, ...,  0.30343812,\n          0.02419355, -0.06909582]],\n\n       [[ 0.10833333, -3.13410771, -3.15672823, ..., -0.6080328 ,\n          1.15322581, -1.03940621],\n        [ 0.12083333, -3.15839493, -3.05329815, ..., -0.47008727,\n          1.15020161, -1.03535762],\n        [ 0.16666667, -3.05807814, -3.08390501, ..., -0.32614867,\n          1.14516129, -1.02887989],\n        ...,\n        [ 0.16666667, -2.93664203, -2.92453826, ...,  0.2736831 ,\n          0.00907258, -0.07287449],\n        [ 0.17916667, -2.9281943 , -2.95831135, ...,  0.30343812,\n          0.02419355, -0.06909582],\n        [ 0.19166667, -2.96092925, -2.92559367, ...,  0.43623173,\n          0.03225806, -0.06396761]],\n\n       [[ 0.12083333, -3.15839493, -3.05329815, ..., -0.47008727,\n          1.15020161, -1.03535762],\n        [ 0.16666667, -3.05807814, -3.08390501, ..., -0.32614867,\n          1.14516129, -1.02887989],\n        [ 0.17916667, -3.0876452 , -3.07862797, ..., -0.33287772,\n          1.1421371 , -1.02132254],\n        ...,\n        [ 0.17916667, -2.9281943 , -2.95831135, ...,  0.30343812,\n          0.02419355, -0.06909582],\n        [ 0.19166667, -2.96092925, -2.92559367, ...,  0.43623173,\n          0.03225806, -0.06396761],\n        [ 0.20416667, -2.92713833, -2.90026385, ...,  0.54620965,\n          0.03629032, -0.05695007]],\n\n       [[ 0.16666667, -3.05807814, -3.08390501, ..., -0.32614867,\n          1.14516129, -1.02887989],\n        [ 0.17916667, -3.0876452 , -3.07862797, ..., -0.33287772,\n          1.1421371 , -1.02132254],\n        [ 0.19166667, -3.07919747, -3.11029024, ..., -0.46083482,\n          1.14314516, -1.01808367],\n        ...,\n        [ 0.19166667, -2.96092925, -2.92559367, ...,  0.43623173,\n          0.03225806, -0.06396761],\n        [ 0.20416667, -2.92713833, -2.90026385, ...,  0.54620965,\n          0.03629032, -0.05695007],\n        [ 0.25      , -2.89968321, -2.97308707, ...,  0.41709599,\n          0.05040323, -0.05668016]],\n\n       [[ 0.17916667, -3.0876452 , -3.07862797, ..., -0.33287772,\n          1.1421371 , -1.02132254],\n        [ 0.19166667, -3.07919747, -3.11029024, ..., -0.46083482,\n          1.14314516, -1.01808367],\n        [ 0.20416667, -3.11298838, -3.11029024, ..., -0.58395542,\n          1.14616935, -1.01403509],\n        ...,\n        [ 0.20416667, -2.92713833, -2.90026385, ...,  0.54620965,\n          0.03629032, -0.05695007],\n        [ 0.25      , -2.89968321, -2.97308707, ...,  0.41709599,\n          0.05040323, -0.05668016],\n        [ 0.2625    , -2.97782471, -3.03430079, ...,  0.28377668,\n          0.07258065, -0.05748988]]])"
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_val[:5]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "outputs": [
    {
     "data": {
      "text/plain": "array([-2.92559367, -2.90026385, -2.97308707, -3.03430079, -3.08073879])"
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y_val[:5]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "outputs": [
    {
     "data": {
      "text/plain": "(66493, 100, 9)"
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "outputs": [
    {
     "data": {
      "text/plain": "(66493,)"
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y_train.shape"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Нейронка"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "source": [
    "from tensorflow.keras.layers import Dense, Dropout, Activation, Flatten, Conv1D, \\\n",
    "                                    LSTM, GRU, Concatenate, Reshape, BatchNormalization, ReLU, \\\n",
    "                                    Average, Multiply, Add\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import Model, Input\n",
    "from tensorflow.keras import layers"
   ],
   "metadata": {
    "id": "BxJ3Q8AtwQL7"
   },
   "execution_count": 3,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "'''\n",
    "activ = 'selu'\n",
    "activ_main = 'tanh'\n",
    "\n",
    "a = Input(shape=(24,24,))\n",
    "\n",
    "z = Conv1D(24, 12)(a)\n",
    "z = Conv1D(24, 12)(z)\n",
    "z = LSTM(24)(z)\n",
    "z = Reshape((6,4))(z)\n",
    "z = GRU(24)(z)\n",
    "#z = Reshape((6,4))(z)\n",
    "#z = LSTM(24)(z)\n",
    "#z = Reshape((6,4))(z)\n",
    "#z = LSTM(24)(z)\n",
    "z = Activation(activ)(z)\n",
    "\n",
    "y = Conv1D(12, 8)(a)\n",
    "y = Conv1D(12, 8)(y)\n",
    "y = LSTM(12)(y)\n",
    "y = Reshape((3,4))(y)\n",
    "y = GRU(12)(y)\n",
    "#y = Reshape((3,4))(y)\n",
    "#y = LSTM(12)(y)\n",
    "#y = Reshape((3,4))(y)\n",
    "#y = LSTM(12)(y)\n",
    "y = Activation(activ)(y)\n",
    "\n",
    "w = Conv1D(6, 3)(a)\n",
    "w = Conv1D(6, 3)(w)\n",
    "w = LSTM(4)(w)\n",
    "w = Reshape((2,2))(w)\n",
    "w = GRU(4)(w)\n",
    "#w = Reshape((2,2))(w)\n",
    "#w = LSTM(4)(w)\n",
    "#w = Reshape((2,2))(w)\n",
    "#w = LSTM(4)(w)\n",
    "w = Activation(activ)(w)\n",
    "\n",
    "x = Concatenate()([z, y, w])\n",
    "\n",
    "x = Dense(1024)(x)\n",
    "x = Dense(128)(x)\n",
    "x = Activation(activ)(x)\n",
    "\n",
    "x = Flatten()(x)\n",
    "\n",
    "x = Dropout(0.3)(x)\n",
    "x = Dense(24)(x)\n",
    "x = Dense(24)(x)\n",
    "\n",
    "x = Activation(activ_main)(x)\n",
    "model = Model(inputs=a, outputs=x)\n",
    "'''"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 145
    },
    "id": "2qi6g6PVwYFy",
    "outputId": "25fa3066-154d-4f8b-ff01-474545aa45ea"
   },
   "execution_count": 19,
   "outputs": [
    {
     "data": {
      "text/plain": "\"\\nactiv = 'selu'\\nactiv_main = 'tanh'\\n\\na = Input(shape=(24,24,))\\n\\nz = Conv1D(24, 12)(a)\\nz = Conv1D(24, 12)(z)\\nz = LSTM(24)(z)\\nz = Reshape((6,4))(z)\\nz = GRU(24)(z)\\n#z = Reshape((6,4))(z)\\n#z = LSTM(24)(z)\\n#z = Reshape((6,4))(z)\\n#z = LSTM(24)(z)\\nz = Activation(activ)(z)\\n\\ny = Conv1D(12, 8)(a)\\ny = Conv1D(12, 8)(y)\\ny = LSTM(12)(y)\\ny = Reshape((3,4))(y)\\ny = GRU(12)(y)\\n#y = Reshape((3,4))(y)\\n#y = LSTM(12)(y)\\n#y = Reshape((3,4))(y)\\n#y = LSTM(12)(y)\\ny = Activation(activ)(y)\\n\\nw = Conv1D(6, 3)(a)\\nw = Conv1D(6, 3)(w)\\nw = LSTM(4)(w)\\nw = Reshape((2,2))(w)\\nw = GRU(4)(w)\\n#w = Reshape((2,2))(w)\\n#w = LSTM(4)(w)\\n#w = Reshape((2,2))(w)\\n#w = LSTM(4)(w)\\nw = Activation(activ)(w)\\n\\nx = Concatenate()([z, y, w])\\n\\nx = Dense(1024)(x)\\nx = Dense(128)(x)\\nx = Activation(activ)(x)\\n\\nx = Flatten()(x)\\n\\nx = Dropout(0.3)(x)\\nx = Dense(24)(x)\\nx = Dense(24)(x)\\n\\nx = Activation(activ_main)(x)\\nmodel = Model(inputs=a, outputs=x)\\n\""
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "activ = 'tanh' # tanh\n",
    "activ_main = 'selu' # sigmoid\n",
    "\n",
    "a = Input(shape=(massiv_blok,9,))\n",
    "\n",
    "'''\n",
    "z = Conv1D(massiv_blok, 8, padding='same')(a) #1\n",
    "#z = Conv1D(massiv_blok, 48, padding='same')(z) #1\n",
    "##z = LSTM(massiv_blok//2)(a)\n",
    "##z = Flatten()(z)\n",
    "#z = Reshape((massiv_blok//2//4,2))(z)\n",
    "#z = GRU(24)(z)\n",
    "#z = Reshape((6,4))(z)\n",
    "#z = LSTM(24)(z)\n",
    "#z = Reshape((6,4))(z)\n",
    "#z = LSTM(24)(z)\n",
    "#z = Flatten()(a)\n",
    "#z = Dense(256)(z)\n",
    "z = Activation(activ)(z) #1\n",
    "\n",
    "y = Conv1D(massiv_blok//2, 16, padding='same')(a)\n",
    "##y = Flatten()(y)\n",
    "#y = Conv1D(massiv_blok, 8, padding='same')(y)\n",
    "#y = LSTM(blok_old//4)(a)\n",
    "#y = Reshape((3,4))(y)\n",
    "#1 y = GRU(192*2)(a) #1\n",
    "#1 y = Reshape((96,4))(y) #1\n",
    "#y = LSTM(12)(y)\n",
    "#y = Reshape((3,4))(y)\n",
    "#y = LSTM(12)(y)\n",
    "##y = Flatten()(a)\n",
    "##y = Dense(1024)(y)\n",
    "y = Activation(activ)(y) #1\n",
    "\n",
    "#k = Flatten()(a)\n",
    "#k = Dense(768)(k)\n",
    "#k = Activation(activ)(k) #2\n",
    "\n",
    "#l = Flatten()(a)\n",
    "#l = Dense(128)(l)\n",
    "#l = Dense(128)(l)\n",
    "#l = Activation(activ)(l) #2\n",
    "\n",
    "w = Conv1D(massiv_blok//4, 32, padding='same', strides=1, kernel_initializer=\"glorot_uniform\")(a) #1\n",
    "#w = Conv1D(massiv_blok//8, 48*16, padding='same', strides=1, kernel_initializer=\"glorot_uniform\")(w) #1\n",
    "##w = LSTM(massiv_blok//8)(a)\n",
    "##w = Flatten()(w)\n",
    "#w = Reshape((massiv_blok//8//2,2))(w)\n",
    "#w = GRU(4)(w)\n",
    "#w = Reshape((2,2))(w)\n",
    "#w = LSTM(4)(w)\n",
    "#w = Reshape((2,2))(w)\n",
    "#w = LSTM(4)(w)\n",
    "#w = Flatten()(w)\n",
    "#w = Dense(512)(w)\n",
    "w = Activation(activ)(w) #1\n",
    "'''\n",
    "l = layers.Bidirectional(layers.LSTM(48, return_sequences=True))(a)\n",
    "l = layers.LSTM(24, return_sequences=True)(l)\n",
    "l = BatchNormalization()(l)\n",
    "\n",
    "\n",
    "n = layers.Bidirectional(layers.LSTM(64, return_sequences=True))(a)\n",
    "n = layers.LSTM(32, return_sequences=True)(n)\n",
    "n = BatchNormalization()(n)\n",
    "\n",
    "#z = BatchNormalization()(z) #1\n",
    "#z = Dropout(0.3)(z)\n",
    "#y = BatchNormalization()(y)\n",
    "#y = Dropout(0.3)(y) #1\n",
    "#w = BatchNormalization()(w) #1\n",
    "#w = Dropout(0.3)(w) #1\n",
    "\n",
    "x = Concatenate()([l, n])\n",
    "#x = Concatenate()([z, y, w, k, l]) #1\n",
    "#x = Concatenate()([z, y, w, n]) #1\n",
    "#x = Flatten()(x)\n",
    "\n",
    "x = Dense(144)(x)\n",
    "x = Activation(activ)(x)\n",
    "x = Dense(48)(x)\n",
    "x = Activation(activ)(x)\n",
    "\n",
    "x = Flatten()(x)\n",
    "\n",
    "#x = Dropout(0.3)(x)\n",
    "#x = Dense(24)(x)\n",
    "x = Dense(1)(x)\n",
    "\n",
    "x = Activation(activ_main)(x)\n",
    "model = Model(inputs=a, outputs=x)"
   ],
   "metadata": {
    "id": "VLTWIYjymDiP"
   },
   "execution_count": 48,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "model.summary()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "doFrdka8yFh_",
    "outputId": "2e920993-1014-420e-9de3-9579819f0d87"
   },
   "execution_count": 49,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_2 (InputLayer)           [(None, 100, 9)]     0           []                               \n",
      "                                                                                                  \n",
      " bidirectional_2 (Bidirectional  (None, 100, 96)     22272       ['input_2[0][0]']                \n",
      " )                                                                                                \n",
      "                                                                                                  \n",
      " bidirectional_3 (Bidirectional  (None, 100, 128)    37888       ['input_2[0][0]']                \n",
      " )                                                                                                \n",
      "                                                                                                  \n",
      " lstm_5 (LSTM)                  (None, 100, 24)      11616       ['bidirectional_2[0][0]']        \n",
      "                                                                                                  \n",
      " lstm_7 (LSTM)                  (None, 100, 32)      20608       ['bidirectional_3[0][0]']        \n",
      "                                                                                                  \n",
      " batch_normalization_3 (BatchNo  (None, 100, 24)     96          ['lstm_5[0][0]']                 \n",
      " rmalization)                                                                                     \n",
      "                                                                                                  \n",
      " batch_normalization_4 (BatchNo  (None, 100, 32)     128         ['lstm_7[0][0]']                 \n",
      " rmalization)                                                                                     \n",
      "                                                                                                  \n",
      " concatenate (Concatenate)      (None, 100, 56)      0           ['batch_normalization_3[0][0]',  \n",
      "                                                                  'batch_normalization_4[0][0]']  \n",
      "                                                                                                  \n",
      " dense (Dense)                  (None, 100, 144)     8208        ['concatenate[0][0]']            \n",
      "                                                                                                  \n",
      " activation (Activation)        (None, 100, 144)     0           ['dense[0][0]']                  \n",
      "                                                                                                  \n",
      " dense_1 (Dense)                (None, 100, 48)      6960        ['activation[0][0]']             \n",
      "                                                                                                  \n",
      " activation_1 (Activation)      (None, 100, 48)      0           ['dense_1[0][0]']                \n",
      "                                                                                                  \n",
      " flatten (Flatten)              (None, 4800)         0           ['activation_1[0][0]']           \n",
      "                                                                                                  \n",
      " dense_2 (Dense)                (None, 1)            4801        ['flatten[0][0]']                \n",
      "                                                                                                  \n",
      " activation_2 (Activation)      (None, 1)            0           ['dense_2[0][0]']                \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 112,577\n",
      "Trainable params: 112,465\n",
      "Non-trainable params: 112\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "keras.utils.plot_model(model, show_shapes=True)"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "BTIGF89WLWN5",
    "outputId": "b4860af0-5700-4a15-93a0-427ce737da5e"
   },
   "execution_count": 65,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "You must install pydot (`pip install pydot`) and install graphviz (see instructions at https://graphviz.gitlab.io/download/) for plot_model/model_to_dot to work.\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "model.compile(loss=keras.losses.logcosh, #BinaryCrossentropy / Hinde /\n",
    "    optimizer=keras.optimizers.Adagrad(), # Ftrl / SGD /\n",
    "    metrics=[keras.metrics.mean_squared_error]) #BinaryAccuracy / BinAcc /"
   ],
   "metadata": {
    "id": "7-AjtHZCwY5y"
   },
   "execution_count": 59,
   "outputs": []
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "outputs": [],
   "source": [
    "# Параметры для функции callback\n",
    "# Прерывание обучение при не изменной ошибке\n",
    "stop = keras.callbacks.EarlyStopping(monitor='val_loss', min_delta=0,\n",
    "                                     patience=17, verbose=1, mode='auto', baseline=None, restore_best_weights=True)\n",
    "# Измененение шага для оптимизатора при стагнации ошибки\n",
    "reduce_lr = keras.callbacks.ReduceLROnPlateau(monitor='val_loss', factor=0.3,\n",
    "                                              patience=11, verbose=1,\n",
    "                                              mode='auto', cooldown=1, min_lr=1e-15)\n",
    "# Сохранение весов, лучшей модели\n",
    "weights_file = 'weights.h5'\n",
    "save_goods = keras.callbacks.ModelCheckpoint(weights_file,\n",
    "                                             monitor='val_mean_squared_error',\n",
    "                                             mode='auto',\n",
    "                                             verbose=1,\n",
    "                                             save_best_only=True)\n",
    "# Коллбэки\n",
    "#pltGraf = LambdaCallback(on_epoch_end=my_graf)  # Конец эпохи\n",
    "#time_start = LambdaCallback(on_epoch_begin=my_time_start)  # Начало эпохи\n",
    "#time_stop = LambdaCallback(on_epoch_end=my_time)  # Начало эпохи\n",
    "#init = LambdaCallback(on_train_begin=my_init)  # Начало обучения, инициализация параметров\n"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "source": [
    "history = model.fit(X_train,Y_train,\n",
    "                    validation_data=(X_test, Y_test),\n",
    "                    batch_size=300,\n",
    "                    epochs=20,\n",
    "                    verbose=1,\n",
    "                    callbacks=[stop, save_goods, reduce_lr])"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "pioWNVHcwc6k",
    "outputId": "9b563b48-044a-407f-d3b2-ae1cbd453795"
   },
   "execution_count": 61,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 1: val_mean_squared_error improved from inf to 0.41978, saving model to weights.h5\n",
      "222/222 [==============================] - 802s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 2/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 2: val_mean_squared_error did not improve from 0.41978\n",
      "222/222 [==============================] - 563s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 3/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 3: val_mean_squared_error did not improve from 0.41978\n",
      "222/222 [==============================] - 543s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 4/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 4: val_mean_squared_error improved from 0.41978 to 0.41977, saving model to weights.h5\n",
      "222/222 [==============================] - 558s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 5/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 5: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 549s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 6/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 6: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 550s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 7/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 7: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 559s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 8/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 8: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 563s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 9/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 9: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 557s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 10/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 10: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 559s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 11/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 11: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 552s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 12/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 12: val_mean_squared_error did not improve from 0.41977\n",
      "\n",
      "Epoch 12: ReduceLROnPlateau reducing learning rate to 0.0003000000142492354.\n",
      "222/222 [==============================] - 556s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 0.0010\n",
      "Epoch 13/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 13: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 556s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 14/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 14: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 559s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 15/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 15: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 550s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 16/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 16: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 552s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 17/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 17: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 564s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 18/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 18: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 560s 3s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 19/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 19: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 552s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n",
      "Epoch 20/20\n",
      "222/222 [==============================] - ETA: 0s - loss: 0.1603 - mean_squared_error: 0.3566\n",
      "Epoch 20: val_mean_squared_error did not improve from 0.41977\n",
      "222/222 [==============================] - 539s 2s/step - loss: 0.1603 - mean_squared_error: 0.3566 - val_loss: 0.1847 - val_mean_squared_error: 0.4198 - lr: 3.0000e-04\n"
     ]
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "model.save('model_fx.h5')"
   ],
   "metadata": {
    "id": "9574nyqT_vyE"
   },
   "execution_count": 62,
   "outputs": []
  },
  {
   "cell_type": "code",
   "source": [
    "metok = len(list(history.history.keys()))//2\n",
    "\n",
    "plt.plot(history.history[list(history.history.keys())[0]])\n",
    "plt.plot(history.history[list(history.history.keys())[metok]])\n",
    "plt.title('model loss')\n",
    "plt.show()\n",
    "for i in range(metok-1):\n",
    "  plt.plot(history.history[list(history.history.keys())[i+1]])\n",
    "  plt.plot(history.history[list(history.history.keys())[i+1+metok]])\n",
    "  plt.title('model metrics')\n",
    "  plt.show()"
   ],
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 545
    },
    "id": "G7ZC-JeG0337",
    "outputId": "c20d67a3-3ee1-45f1-ef9c-66c321dfbcd7"
   },
   "execution_count": 63,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ]
  },
  {
   "cell_type": "code",
   "source": [
    "from keras.models import load_model\n",
    "model = load_model('model_fx.h5')"
   ],
   "metadata": {
    "id": "uk_xU74f9tjY"
   },
   "execution_count": 64,
   "outputs": []
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "outputs": [
    {
     "data": {
      "text/plain": "[<matplotlib.lines.Line2D at 0x17ba67716c0>]"
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ttt = 100\n",
    "val_predict = model.predict(X_val[ttt:ttt+massiv_blok])\n",
    "val_original = Y_val[ttt:ttt+massiv_blok]\n",
    "plt.plot(val_predict[:,0] - val_original)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "outputs": [
    {
     "data": {
      "text/plain": "array([[0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442],\n       [0.05382442]], dtype=float32)"
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_predict"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "outputs": [
    {
     "data": {
      "text/plain": "array([-2.66490765, -2.59313984, -2.53720317, -2.64379947, -2.72295515,\n       -2.66279683, -2.63852243, -2.61002639, -2.57730871, -2.52031662,\n       -2.54881266, -2.57836412, -2.44010554, -2.42427441, -2.46965699,\n       -2.49815303, -2.53192612, -2.49182058, -2.47598945, -2.5055409 ,\n       -2.56781003, -2.5298153 , -2.51292876, -2.50976253, -2.49604222,\n       -2.48970976, -2.46437995, -2.46754617, -2.41477573, -2.46965699,\n       -2.50976253, -2.48548813, -2.48865435, -2.43060686, -2.46121372,\n       -2.66807388, -2.61002639, -2.74722955, -2.48443272, -2.40211082,\n       -2.20369393, -2.15936675, -2.18575198, -2.23007916, -2.12664908,\n       -2.33562005, -2.5762533 , -2.54986807, -2.43377309, -2.65646438,\n       -2.62902375, -2.65963061, -2.67757256, -2.67440633, -2.43588391,\n       -2.42744063, -2.28073879, -2.36411609, -2.40105541, -2.37044855,\n       -2.33034301, -2.29023747, -2.52031662, -2.54881266, -2.43377309,\n       -2.40316623, -2.41266491, -2.38100264, -2.36622691, -2.33034301,\n       -2.42744063, -2.45065963, -2.50765172, -2.49076517, -2.56042216,\n       -2.46860158, -2.38522427, -2.21952507, -2.1646438 , -2.30923483,\n       -2.36622691, -2.40949868, -2.40738786, -2.47598945, -2.52875989,\n       -2.56358839, -2.46860158, -2.52348285, -2.43271768, -2.50131926,\n       -2.38627968, -2.51609499, -2.50976253, -2.54775726, -2.52137203,\n       -2.39050132, -2.46965699, -2.45488127, -2.52031662, -2.4348285 ])"
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_original"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "outputs": [],
   "source": [],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "8a361aba61e6869f933fe5c4d85d9bd0f469a48fae73d03aae6deed734b848e5"
  },
  "kernelspec": {
   "display_name": "Python 3.9.5 64-bit",
   "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": "3.9.5"
  },
  "orig_nbformat": 4,
  "colab": {
   "name": "12-24.ipynb",
   "provenance": [],
   "collapsed_sections": []
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}