RedesRecorrentes_DoencasPulmonares_LSTM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "RedesRecorrentes_DoencasPulmonares_LSTM.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3",
      "language": "python"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/rrfsantos/Projeto-Rede-Neurais--Series-Temporais/blob/main/RedesRecorrentes_DoencasPulmonares_LSTM.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bgc8yb4c2W0g"
      },
      "source": [
        "# Recurrent Neural Network - LSTM\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aew2N81e35Uc"
      },
      "source": [
        "## Parte 1 - Preprocessamento de Dados"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z0vulSxH2hHu"
      },
      "source": [
        "### Importar as bibliotecas"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p2SjiqlE2O08"
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import os\n",
        "import math\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense\n",
        "from keras.layers import LSTM, GRU\n",
        "from keras.layers import Dropout\n",
        "from sklearn.metrics import mean_absolute_error, mean_squared_error\n",
        "np.random.seed(0)"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "36Q16Udt2l3W"
      },
      "source": [
        "### Importar a base de dados"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "or7UuAX_PgCb",
        "outputId": "2df7da55-baee-4a78-b75a-2ae7fd751ec4"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "\n",
        "import os\n",
        "workdir_path = '/content/drive/My Drive/Trabalho Redes Neurais/'  # Inserir o local da pasta onde estão os arquivos de entrada (treino e teste)\n",
        "os.chdir(workdir_path)"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1ivhnWxD2eEG"
      },
      "source": [
        "dataset_train = pd.read_csv('train.csv')\n",
        "training_set = dataset_train.iloc[:, 1:2].values #só coluna da série em si"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kd39T1zK2ql1"
      },
      "source": [
        "### gráfico da série temporal"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vMsAFuya2rhr",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "410b5f09-8cc7-4194-d31d-62a940f00439"
      },
      "source": [
        "plt.plot(training_set)\n",
        "plt.xlabel(\"Tempo\")\n",
        "plt.ylabel(\"Numero de Mortes\")\n",
        "plt.title(\"Doenças Pulmonares\")\n",
        "plt.show()"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOydeXxcdbn/309mJjNZJvukWdt0X2lLKVBANpVFQMAN9bqBCJcrLle9Kl7Xq+K9Xq/LxR8uKAiKiChXrYiyCWIR6EZbaEv3tGmbZl8m22SW7++Pc04ypFkm6ZyZM8n3/XrNq5OzzXfSzDzn2T6PKKXQaDQajWY8stK9AI1Go9E4H20sNBqNRjMh2lhoNBqNZkK0sdBoNBrNhGhjodFoNJoJ0cZCo9FoNBOijYVG4zBE5F4R+Xq616HRxKONhSZtiEi9iPSLSFBEOkXkHyJyi4hk/N+liNSJiBKRHvNRLyK3pXtdGs1Ucad7AZoZz5uVUk+KSCFwIfC/wNnADeldVtIoUkpFROQc4CkR2aaU+ku6F3WqiIhbKRVJ9zo0qSPj7+A00wOlVJdSaj3wTuADIrICQEQKReTnItIiIodF5AvxnoeIfFBEdotIh4g8JiJz4vYp01PZZ3oud4qIxO2/yTw3KCK7RGSNuf02ETkQt/0tcecsEJG/iUiXiLSKyK8TfH/PAzuBFSJyvYhsiN9vrnXByPNE5CIROSoinxGRZhFpFJFrReQKEdkrIu0i8u9xx3tF5Hsictx8fE9EvCOu9am4a90Qd+6VIvKSiHSLSIOIfCVun+Up3SgiR4C/jvf7F4Pvmq/TLSIvW/+nmsxEGwuNo1BKbQSOAuebm74PFALzMDyP92N6HSJyDfDvwFuBAPB34FcjLnkVcCawErgOuMw89x3AV8zrFQBXA23mOQfM1y8E/gO4X0QqzX1fAx4HioEac33jYn5xngcsB15K5PcwggrAB1QDXwJ+ArwXOMNc5xdFZK557OeBdcBqYBVwFvCFEdcqNK91I3CniBSb+3oxfh9FwJXAv4jItSPWciGwFLhsgt//pcAFwCLz9a5j+PeryUSUUvqhH2l5APXAG0fZ/gLGl54LGASWxe37Z+AZ8/mfgRvj9mUBfcAc82cFvC5u/0PAbebzx4CPJ7jObcA15vOfA3cBNROcU2e+fifQAewGPmbuux7YMOJ4BSwwn98LfN18fhHQD7jMn/3msWfHnbsFuNZ8fgC4Im7fZUD9iGu54/Y3A+vGeA/fA7474v3Mi9s/5u8feD2wF8NwZaX7b00/Tv2hPQuNE6kG2oEywAMcjtt32NwPxpfS/5ohpk7zHInbD3Ai7nkfkG8+r8X4Yj0JEXm/iGyLu+4Kcy0AnzFfY6OI7BSRD07wXsqUUsVKqaVKqTsmOHYs2pRSUfN5v/lvU9z+fobfVxUn/76qRlwrPtcw9DsRkbNF5Gkz5NcF3MLw+7ZoiHs+5u9fKfVX4P8BdwLNInKXiBQk/pY1TkMbC42jEJEzMb7sNwCtQBjjS8liNnDMfN4A/LNSqijukaOU+kcCL9UAzB/l9edghHk+ApQqpYqAVzC+BFFKnVBK3aSUqsLwcn4wWq5hAnqB3LjXrJjk+eNxnJN/X8cTPPcBYD1Qq5QqBH6E+b7jiJepHvf3r5S6Qyl1BrAMIxz16cm/HY1T0MZC4whEpEBErgIeBO5XSr1s3k0/BNwuIn7zi/yTwP3maT8CPiciy81rFJq5iET4KfBvInKGmVNYYF4/D+MLscW85g0YnoW1zneISI35Y4d5bGySb3c7sFxEVouIDyN3kix+BXxBRAIiUoaR47h/gnMs/EC7UmpARM4C/mmC48f8/YvImaan4sEwjgNM/vekcRDaWGjSzR9FJIhxl/p54Du8tmz2oxhfNgcxvI0HgHsAlFK/A74JPCgi3RgewJsSeVGl1G+A283rxYDfAyVKqV3At4HnMUI9pwHPxZ16JvCiiPRg3IV/XCl1cDJvWCm1F/gq8CSwz3xfyeLrwGZgB/AysNXclggfBr5q/n98CcNQj8kEv/8CDA+tAyMU1gZ8a1LvROMoRCk9/EgzsxGRHwPfNr/ENRrNKGjPQjOjEZF8jJj+Belei0bjZHQHt2amcwAjDPWGdC9Eo3EyOgyl0Wg0mgmxPQwlIi5TQuAR8+e5IvKiiOwXkV+LSLa53Wv+vN/cXxd3jc+Z2/eIyGV2r1mj0Wg0ryUVYaiPY3SvWg0538ToCn1QRH6EITnwQ/PfDqXUAhF5l3ncO0VkGfAuDKmEKuBJEVkU16R0EmVlZaqurs62N6TRaDTTkS1btrQqpQKj7bPVWJj16FdilCh+UkQEQwbAqt++D6PG/IfANQzXm/8W+H/m8dcADyqlQsAhEdmPoXfz/FivW1dXx+bNm5P+fjQajWY6IyKHx9pndxjqexjyCFYzTinQGSc3cJRhaYZqTCkBc3+XefzQ9lHOGUJEbhaRzSKyuaWlJdnvQ6PRaGY0thkLsxu3WSm1xa7XiEcpdZdSaq1Sam0gMKoXpdFoNJopYmcY6jzgahG5AkNeuQBjsE2RDA9OqWFY5+cYhrjbURFxY8gat8Vtt4g/R6PRaDQpwDbPQin1OaVUjVKqDiNB/Vel1HuAp4G3m4d9APiD+Xy9+TPm/r8qo653PfAus1pqLrAQ2GjXujUajUZzMuloyvsshpbM1zEGwdxtbr8b+IWZwG7HMDAopXaKyEPALiAC3DpeJZRGo9Foks+0bMpbu3at0tVQGo1GMzlEZItSau1o+7Q2lEaj0WgmRBsLjWYEv3vpKMGBcLqXodE4Cm0sNJo4Gtr7+MSvt/OHbYkOl9NoZgbaWGg0cXT1Gx5Fc/dAmlei0TgLbSw0mji6TWPR0hNK80o0GmehjYVGE0f3gKFE09ytjYVGE482FhpNHN0D2rPQaEZDGwuNJo6hMFRQGwuNJh5tLDSaOIJmGKolGCIWm34NqxrNVNHGQqOJwwpDRWKKzn7da6HRWGhjodHE0d0fGXquQ1EazTDaWGg0ccR3bmtjodEMo42FRhNH90CYgN8LQHNQN+ZpNBbaWGg0cXT3R5gfyAO0Z6HRxKONhUYTR/dAmMrCHHI8Lm0sNJo4tLHQaOIIDkQo8LkJ+L00a2Oh0QyhjYVGYxKLKYIDYQpyPJT7vdqz0Gji0MZCozHpHYwQU1Dg8xDwe7Xkh0YThzYWGo2J1b3tt8JQWqZcoxlCGwuNxsTq3rbCUN0DEQbC0TSvSqNxBtpYaDQmVve2FYYCaNWhKI0G0MZCoxnC6t62wlCArojSaEzc6V6ARuMU4sNQriwBdGOeRmOhjYVGYzIchnKTm+0CtLHQaCy0sdBoTIbDUB6yBES0sdBoLLSx0GhMugci+DxZZLuNVF5pXrbOWWg0JjrBrdGYdPeHKfB5hn4uy9dd3BqNhTYWGo1JcCCC3zfsbOsubo1mGG0sNBqTblMXyqLc76NFd3FrNICNxkJEfCKyUUS2i8hOEfkPc/u9InJIRLaZj9XmdhGRO0Rkv4jsEJE1cdf6gIjsMx8fsGvNmpnNyDCU5VkopdK4Ko3GGdiZ4A4Br1dK9YiIB9ggIn82931aKfXbEce/CVhoPs4GfgicLSIlwJeBtYACtojIeqVUh41r18xAggMRaktyh34O+L2Eo4qu/jBFudlpXJlGk35s8yyUQY/5o8d8jHeLdg3wc/O8F4AiEakELgOeUEq1mwbiCeByu9atmbmcHIbSXdwajYWtOQsRcYnINqAZ4wv/RXPX7Wao6bsi4jW3VQMNcacfNbeNtX3ka90sIptFZHNLS0vS34tm+tPdHzkpDAW610KjAZuNhVIqqpRaDdQAZ4nICuBzwBLgTKAE+GySXusupdRapdTaQCCQjEtqZhAD4SiD0dhJ1VCgjYVGAymqhlJKdQJPA5crpRrNUFMI+BlwlnnYMaA27rQac9tY2zWapBGvC2UxLCaoK6I0GjuroQIiUmQ+zwEuAV418xCIiADXAq+Yp6wH3m9WRa0DupRSjcBjwKUiUiwixcCl5jaNJmnE60JZ+L1ufJ4s7VloNNhbDVUJ3CciLgyj9JBS6hER+auIBAABtgG3mMc/ClwB7Af6gBsAlFLtIvI1YJN53FeVUu02rlszA7F0oeJzFiJilM9qY6HR2GcslFI7gNNH2f76MY5XwK1j7LsHuCepC9Ro4ug2R6oW5Lz2IxHI9+pqKI0G3cGt0QBGQx681rMAs4tbGwuNRhsLjQaGE9z+EcZC60NpNAbaWGg0GN3bMEoYyu+lsy9MKBJNx7I0GsegjYVGgxGGcmcJOR7Xa7ZbXdytPYPpWJZG4xi0sdBoMMJQfp8bo6J7GN2Yp9EYaGOh0WCEoeIb8iy0sdBoDLSx0Gg4WZ7cotzvA3QXt0ajjYVGg9FnEa8LZVGab0iTa89CM9PRxkKjwejgHs2z8LiyKMnL1sZCM+PRxkKjwZQnzxld0KDcr7u4NRptLDQarGqokz0LQOtDaTRoY6HREInG6BuMjhqGAkMfShsLzUxHGwvNjGes7m2LQIFhLAytS41mZqKNhWbGM5YulEUg38tgNDY080KjmYloY6GZ8Qx5FqOUzkJcY16P7rXQzFy0sdDMeIbkyUfp4Ia48ardOm+hmbloY6GZ8QyHocYqnTW6uLVUuWYmo42FZsYzNCVvnNJZ0F3cmpmNNhaaGc9EYagCn5tsd5Y2FpoZzYTGQkTeISJ+8/kXROT/RGSN/UvTaFJD90AEEfB7Rw9DiQgFPs9QuEqjmYkk4ll8USkVFJHXAW8E7gZ+aO+yNJrUERwIk5/tJitLxjymwOceqprSaGYiiRgLa57klcBdSqk/Adn2LUmjSS2GLtToISgLvzYW05Jdx7u54L+fprlbl0VPRCLG4piI/Bh4J/CoiHgTPE+jyQisKXnj4fd5COow1LTj6T3NHGnvY1tDZ7qX4ngS+dK/DngMuEwp1QmUAJ+2dVUaTQoZS548nnzv9PEsojHFzuNd6V6GI7CMxKHW3jSvxPlMaCyUUn1AM/A6c1ME2GfnojSaVDKePLnFdApD/XZLA1d9fwPHO/vTvZS0opTSxmISJFIN9WXgs8DnzE0e4H47F6XRpJLuBDyL6RSGem5/G0pBY9fMjtOf6B4YKoc+qI3FhCQShnoLcDXQC6CUOg747VyURpNKgmOMVI3H73PTOxglGsts5VmlFBsPtQPQ3juY5tWkl+0NRihuYXk+9dpYTEgixmJQGdrMCkBE8uxdkkaTOmIxZeQsEqiGAugJZXYo6kh7HyfMyp+2GS5fsv1oJ+4s4YrTKmkOhjL+/9ZuEjEWD5nVUEUichPwJPBTe5eVHsLRGDuPd9EcnNnu+UyidzBCTI0t9WFhGYtMD0W9aHoVAG0z3rPoZGllAUsrjUCJ9i7GJ5EE9/8AvwUeBhYDX1JK3THReSLiE5GNIrJdRHaKyH+Y2+eKyIsisl9Efi0i2eZ2r/nzfnN/Xdy1Pmdu3yMil03trU5MR+8gV96xgb+8csKul9A4DEsXKpHSWSDjk9wbD7VTnOshN9tFW8/MNRaxmOLlo12sqi1kblk+oPMWE5FIgvubSqknlFKfVkr9m1LqCRH5ZgLXDgGvV0qtAlYDl4vIOuCbwHeVUguADuBG8/gbgQ5z+3fN4xCRZcC7gOXA5cAPRMQ1ubeZGAG/lxyPi/rWPjsur3EglqeQaBhqOhiLs+aWUJqfTXvvzA1DHWztJRiKsKqmiDmluQAcatHGYjwSCUNdMsq2N010kjLoMX/0mA8FvB7DUwG4D7jWfH6N+TPm/jeIiJjbH1RKhZRSh4D9wFkJrHvSiAhzSnM53Kb/aGYK1vS7RKqhAHpCmRuGauzq50h7H2fNLaUkzzujw1DbzZLZVbVF+DwuqotyqNef+3EZ01iIyL+IyMvAEhHZEfc4BOxI5OIi4hKRbRh9Gk8AB4BOpZR1e3YUqDafVwMNAOb+LqA0fvso58S/1s0isllENre0tCSyvFGZU5qr/2hmEJbibCLVUJDZnoVVBXX23BLK8rJndBhq+9FO8rJdzA8YIai5ZXk6DDUB43kWDwBvBv5g/ms9zlBKvTeRiyulokqp1UANhjew5NSWO+5r3aWUWquUWhsIBKZ8nbrSPBra+zO+RFKTGMHQ5MJQ3RluLPK9bpZWFlCSlz2jS2e3N3RyWk0hLlM8cm5ZHodaejAKPzWjMaaxUEp1YdzRn66UOhz3aB/rnHGu1Qk8DZyDUVVl3cbVAMfM58eAWgBzfyHQFr99lHOSzpzSPAajsaHyQs30ZjgMNYFn4bUS3Jkbhtp4qJ21dcW4soTSfC9tvaEZ+eUYikTZ3RhkVW3R0La6sjy6ByIz2oBOxLg5C6VUFNgjIrMne2ERCYhIkfk8ByP3sRvDaLzdPOwDGJ4LwHrzZ8z9fzX7O9YD7zKrpeYCC4GNk11PotSZya7D2iWdEQyHocb3LHyeLNxZkrFhqLaeEPuaezhrbgkApXnZhKOK4AzsLXi1MchgNMaqmmFjMa/MaB/Tsh9jk0iCuxjYKSJPich665HAeZXA0yKyA9gEPKGUegRDOuSTIrIfIydxt3n83UCpuf2TwG0ASqmdwEPALuAvwK2mEbOFOeYfTX3bzKmIUkrx9KvNdPWPf9fc3jvIbQ/v4I/bj6doZfYTDEXwebLIdo//URARUx8qMz2LTfUdgJGvACjJM6YMtM/AvMX2o8PJbYu52lhMyPi+t8EXp3JhpdQO4PRRth9klGompdQA8I4xrnU7cPtU1jFZKgp8ZLuyZlRF1MvHurjh3k2U+718/doVXLq84qRj/rG/lX/99TaagyEaOvp486qqNKw0+XT3T6wLZeH3eejJUM9i46F2vO4sTqs2viBL8w1j0dYboq5sZokybGvopCzfS1Whb2hbTXEO7izRxmIcEmnK+xvwKoYelB/YbW6blriyhNqSmVVG9+qJIAA52S5u/sUWbn1g65DAWjga45t/eZX33P0i+T43Z9YVc3Aa1aMnMsvCIpOVZzfWt7FmdvGQB1Wa5wWYkRVRO452sbq2EKMy38DtymJ2aa42FuOQSFPedRg5gndgzLZ4UUTePv5ZmU1daR6HZ1AY6kBzD9muLB771wv4t0sX8cTOJi757t/4+fP1vP1Hz/PDZw7wzrW1PPLR13HhogCNXQP0TpNYd3Bg4il5Fpk606J7IMyu491D+QqI9yxmlrHoHghzoKWHlXH5Cou5pXnaWIxDIrdUnwfOVEo1g5G4xtCH+u24Z2Uwc0rz+MeBNpRSr7n7mK7sb+5hXiAPn8fFR16/kMtXVPCZ3+7gS3/Yid/n5s5/WsOVKysBmGfWpR9q7WVFdWE6l50UuvvDFOUmNiXY7/NwtCPzbiK2HO4gpobzFRCXs5hhxuKVo10o9dp8hcXcsjyeO9BKLKbGncc+U0nEWGRZhsKkjWk+VrWuLJf+cJSWYIjyAt/EJ2Q4+1t6XvPFv6Dcz29uOZfHdp5gVW0R1UU5Q/usJqYDI87JVLoHItSW5CZ0bEGGhqE2HmrHnSWcPrt4aJvP4yLf66Z1hinPbrOS2zUn/+3ODeQxEDbK5qvi/uY1Bol86f9FRB4TketF5HrgT8Cj9i4rvcwpnTkVUQPhKA3tfSwwjYCFy5Rurh7xoZlTmosI0yZvkYg8uYXf585IGeuNh9pZWVNITvZrJdVmYmPejoYu5pTmjupN6oqo8Ukkwf1p4C5gpfm4Syn1WbsXlk6sXouZkOQ+2NJLTMGC8vyJD8a4I60pzsk4aQSl1EkNaEopY6TqZKqhQpGMamTrH4yy42gnZ80tPWnfTDQW2492vqa/Ip55Wn12XBIKJymlHlZKfdJ8/M7uRaWbqqIcXFkyI8pn97cYWo+JGgswPlQHmnsmPtBBPLaziWVfeoz7Xzg89GUfisQYjMYmVQ0VjSn6Bm1r80k6Lx3pIBxVr8lXWJTlZ9M6g6qhmroHaOwaGDVfATCrwFCd1uqzozOekGBQRLpHeQRFpDuVi0w1HlcWNcU5M6Iian9zD1ky7IInwvxAPodae4llkH7WK8e66A9H+cLvX+HWB7bS1R+mO0F5cov8DBQT3HK4AxE4o674pH2GZzFzchaW0uzq2tFzbSJCXVnejIgoTIXxPIunMLqmvw6cppQqMB9+pVRBapaXPubMkPLZA8091Jbk4vMkPiJkXiCP/nA0o/Sz2npDlOVn89nLl/DYziau+v7f2bCvFZhYF8pieABS5nRxH+8aoDQve9RQW2m+l/bewYwKq50KuxuNfqJllWMXZswr0+WzYzGekOC1wGVAC3CXiPxNRD4sIif7s9OQOlOqfLp/kPY395yU3J6IeQHDC8mkJHdLcJCyfC//ctF8HvrndcRi8MmHtgMTz7KwyETl2fbe0FAD3kgsfahMej+nwpH2PioKfCcl+uOZW5bHkfY+wtFYCleWGUwkJNillPoZxrCjHwNfBa5PwbrSzpzSPIIDETr6MucucrJEojEOtfZOKl8Bry2fzRRae0IE/MaX5hlzSvjTx17HpctmIWJIPSSC5YFkUkVUW8/gUE/FSIYa82ZI+WxDRx+1JeP/X88tyyMaUzS0T/+owmQZ1/8WkXOBdwPnAxuAtyil/p6KhaWb+IqosT5smU5DRz+D0RjzJ2ksyv1e8r1uDmaYsbD+TwGKcrP58fvOmFQvTSaGodp7B1laNXrUuMT0ONp7B5k39REwGcPR9j7WzTu5Kiyeurjy2XmT9LinO2MaCxGpBzqBB4GbgYi5fQ2AUmprCtaXNqxei8NtvayZfXJycDqw36xoWjhJYyEizAtk1mSxth4jDBWPiEyq6TLfm3kJ7rbeQcrG8izyZo7kx2AkRmP3ADUTNGBqqfKxGc+zqMeYmX0ZcCkQ3/9uzdKettSW5CDCtE5yW8Zisp4FGB8qS/ba6fSGIvSHo5T5R4/dJ8rwaNXM8CzC0Rhd/eEhD2Ikw2Go6W8sjnf2oxTUThByLM7LpijXo43FKIxpLJRSF6VwHY7D63ZRVTi9y2f3NQeZVeBNOMEbz7xAPr/fdpy+wQi52YlVE6ULS9JipGcxWfKy3YhkjmfRYXoMJfmjexbD+lDTP2dxxMxBJCLtMldXRI3KtNZ4OlXmmBVR05UDzT2TTm5bWBVRmfChshrPSsf40kyUrCzJKOVZK7w0VhjK63bh97pnRGNegykAOVsbiymjjcU4jNdrcdvDO/jSH15J8YqSh1KKAy29ky6btbAqojKhfNbyLAKn6FmAUWabMcbCNALjFWiU5M8MyY+G9n48LmFWAjmquaV5NHYN0J9BnfqpQBuLcagrzaW9d/CkcaP7m3t4cFMD979wOGMlQU50D9ATikzZs5hblpcxgoLJCkOBNdMiM3IWbWZ4aTyPqnSG6EM1dPRRbcr4TMTcgCUk6vy/7VSSyPAjEZH3isiXzJ9ni8hJY1GnI1ZF1JER3sU9zx0i252FK0u4Z8OhdCztlDmV5DYYgoJVhTkZ0WvRGpz4DjtRMmlanuVZjNWUB0b57EyQKT/a3pewFH1daeaEWFNJIp7FD4BzMPotAILAnbatyEHUlRl/XIfbh/9o2nsHeXjLUd62ppqrV1Xz0OajdPZl3p2ZZSym6lkAZvms841FW2+IwhzP0EjRU8HvcxMMZYZn0d47iCtLKBxH+2rmeBb91BQnZiysWRZNGSRnkwoS+fScrZS6FRgAUEp1ANOzS20EVjIsPm/xyxcOE4rE+OB5c/nQ+XPpD0f55YtH0rXEKbO/uYcCn/uU4vjzA/kcanG+JEprj6ELlQz8mZSz6B2kODd73KlvpWbOwun/h6dCTyhCe+9gQsltgKIcDx6X0Byc/h7XZEjEWIRFxIXRW2GNVZ0Rwim52W7K/V7qTXc0FIny8xcOc+GiAAtn+VlaWcD5C8u49x/1hCKZlQzbb1ZCncrY2PmBPHoHozR1O/tD1Ro8uSFvqvh9bnoyxVj0hIYa78aiJC+bSMyY6zFdaRgqm01M1iUrSyjL99KijcVrSMRY3AH8DigXkdsxZD++YeuqHERdXEXUH7c30hIM8aHz5w7tv+n8ebQEQ6zfdjxdS5wS+0+hbNZiXoZoRLX2hE65Ic8ikzyL9t6xdaEsLCPaOo17LYaMRYJhKICAXxuLkSQyKe+XwGeA/wQagWuVUr+xe2FOYU6c+uxP/36QxbP8vG5B2dD+8xeWsaTCz0//fihjXPmO3kHaegdZWO4/pesMq89mgLFIkr6X3+dmMBpjIOx8T7K9d3DC3pLhxrzpm7do6OgHEmvIsyj3e3UYagTjDT8qsR5AM/Ar4AGgaabIlIMhLNYcDPHk7mZePRHkxvPnviZ0IyJ86Px57GkK8qw5H8HpTGU63mhUFPjIzXZxwMHls6FIlO6BSFLDUJAZXdytCYShZoLkR0N7H3nZLopzE1cq0J7FyYznWWwBNpv/tgB7gX3m8y32L80ZzDGVSr/+p12U5Xu5ZnXVScdcvaqKcr+Xn/79YKqXNyWSUQkFmSEoaH0JJi8MlRn6UOFojO6ByJi6UBZWWW3bNA5DHe0wymYnk58L+H2094aIZtA0SLsZb/jRXKXUPOBJ4M1KqTKlVClwFfB4qhaYbuqG1Gf7eP85c/C6Tx6cku3O4vrz6vj7vlZ2HXf+xNn9zT34PFlUFyWW8BsPp8/jTmZDHoDfa9ydOn2mhaULNVEYqjjPeD/t09izODKJHguLgN9LTM2cWR+JkEiCe51S6lHrB6XUn4Fz7VuSs5htehbZ7izec/bsMY97z1lzyM12ZYR3sb+5h3ll+eOWVCbKvEAex7v6HRvDb0uSLpRFpoShhvSwJghDed0u/D73tJUpV0rR0N4/qeQ2DEvD6LzFMIkYi+Mi8gURqTMfnwcyq/TnFCjweagrzeXdZ9ZSOs7daWGuhytPq+SpV5tTuLqpkYxKKIv5gXyUcm63a0sSdaEA8jMkDGUlrBPpWi/Ny562xqKtd5D+cDThslmL8gLj70XnLYZJxFi8GwhglM/+n/n83eOeAYhIrYg8LSK7RGSniHzc3P4VETkmItvMxxVx53xORPaLyB4RuSxu++Xmtv0icttk3+Sp8ujHz+eLVy2b8LgF5UxF1F0AACAASURBVPl09YfpdvAXSd9ghGOd/UkzFk6fx53sMJQl5+70udXDulATv+/SfO+0lSmfStksDN9caGMxzISDCJRS7cDHp3DtCPAppdRWEfEDW0TkCXPfd5VS/xN/sIgsA94FLAeqgCdFZJG5+07gEuAosElE1iuldk1hTVMi0XkNlpzAsY5+CionPyMiFRxoNr7Uk2Us5pqTxZzaa9EaHCQv20VO9sm5pqmQKWGotgTDUGB4H9N15vRUymaBoXntLTpnMYRtqrNKqUZr9KpSKgjsBqrHOeUa4EGlVEgpdQjYD5xlPvYrpQ4qpQYxxrxeY9e6T4UacwrXUfMP1InsaQoCkx+lOha52W6qCn2O7bVo6w0ldHedKMOjVZ3rPUJiulAWZfnTNwxlGcGaCSbkjcTncVHgc9Os9aGGSIlEuYjUAacDL5qbPiIiO0TkHhGxBlxXAw1xpx01t421feRr3Cwim0Vkc0tLS5LfQWIMGwvn3qVta+gg3+tO6jD6eYF8x+YskqkLBeB2ZZGb7XK85Edbb2hCXSiLElNMMDYNy0Qb2vsoy88mzzv5aY4BvzfjPIvvPL6Hrz9iT9DFdmMhIvnAw8C/KqW6gR8C84HVGB3h307G6yil7lJKrVVKrQ0EAsm45KQpycsmx+NytGexraGTlTWFCen6J0p1UQ6NXc68A0umLpRFJsiUt/UMJhSCAkOmPBpTjs61TZWGjr6E1WZHUu730exw3bORvHConZePddly7UTmWdSIyO9EpEVEmkXkYRGpSeTiIuLBMBS/VEr9H4BSqkkpFVVKxYCfYISZAI4BtXGn15jbxtruOESEmuIcx3oWA+EorzYGWV1blNTrzir00dITIhx1nr5kMnWhLPK9zpcpT0Tqw8LyvKbjeNWG9v5J5yssMtGz6OoLUzSJTvXJkIhn8TNgPVCJkXj+o7ltXMRol7wb2K2U+k7c9sq4w94CWLNJ1wPvEhGviMwFFgIbgU3AQhGZKyLZGEnw9QmsOy0YxsKZnsUrx7qIxFTSjUVloQ+lnFeTHo0p2vsGk6YLZZEJYoJtCYgIWkxWH2ogHM2IkFU0pjje2U/tJPMVFpko+dHZP0hRjj0TJBIxFgGl1M+UUhHzcS9G+exEnAe8D3j9iDLZ/xaRl0VkB3Ax8AkApdRO4CFgF/AX4FbTA4kAHwEew0iSP2Qe60hqinMdayxeOtIJwOrZyTUWFeZc4xMOC0UZcxqSJ/Vh4fe5nV86m4AulIUl+ZFI+exLRzp43Tf/yrce33NK60sFjV39RGJqyp5Fud9L32DU8d368XTa6FkkkvVpE5H3YggJgtFj0TbRSUqpDcBogfFHR9lmnXM7cPso2x8d7zwnUVuSM9RrYdXkO4VtDZ1UF+VQ7p94aP1kqCh0prFIdo+FRYHPw/FOZ94QAAxGDF2oRKvAShMMQz228wQff/AlBsIx9plVdU6mod34P0p06NFIhspng6GhKjgn0z8YJRSJUZSbPs/ig8B1wAmMhPTbgRtsWc00IL7Xwmlsa+hMulcBRhgK4ITDygwtY5HoHXaiOD3B3dE3uZnjxbkTh6Hu2XCIW+7fwpKKAlbVFjku5DgaU23Is7BuqjKlfLaz3/j/S0vOwpyQ9w2l1NVKqYBSqlwpda1SKvPmiKYIp/ZaNHcPcKyzn9OTnK8AKMzx4HVncaLLWe852YqzFvleZxuLyTTkgaF7VuBzjyqaF40pvvrHXXz1kV1cumwWv7ppHQvL8zOiSqiho48sgcqiqXnSmdaY19lnFF0UJdBbMxXG9a2UUlERmSMi2WZDnGYCLM/CaRVRLzUY+YrTbfAsRITKQp/jymftCkP5fR76w1HC0RgeV0palSbFZKQ+LErzvSc15kWiMT76q5f48ysnuOG8Or5w5TJcWUK530trT4hYTCVFjNIuGtr7qCzMmfL/UXwYKhOwjEVhGnMWB4HnRGQ9MNR5FV/hpBmmONdDbrbzei22NXTizhKWVxXacv1ZBT6aHOaut/SEyHYZd83JxJL86BmIUJzkEFcymIyIoEWp2ZgXzx1/3c+fXznBv1+xhJsvmD+0vdzvJWJVmiXZECeTho7+SQsIxlOU48HjkowIuQF0WWGoNFZDHQAeMY/1xz00o+DUXouXjnSwrKoAnyc5GkkjcaJn0dZj9BpMZuhNIjhdH2oo/DaJzvWSvOzXTMt7bn8r3//rPt5+Rs1rDAVAuVn95vQ77ob2vinnKwCysoSy/Mwpnx0KQ6XLs1BK/QeAiOQqpZz1DehQaopzhyoxnEA0pnj5aBdvOyOhXsopUVGYQ1N3o6NCE4bUR/LvfP1mlZtTG/PaekO4smRS1Xil+V62mqXVLcEQH39wG/PK8vjqNctPOrbcPzzrYWnlSbsdwUA4SnMwNOVKKItABs3i7uy311gk0sF9jojsAl41f14lIj+wZTXTBKd5Fvuag/QORpPejBdPRYGXcNQITTiFZOtCWRQ43LNo7x1MWBfKojQvm46+QSLRGJ/49TaCA2HufM+aURWXM6FKyPr8TbXHwqI8gxrzOvvCZLuyyLEpepBIGOp7wGWYvRVKqe3ABbasZppQU5xD90CErn5n3HlazXinzy6e4MipU1FoxIad1Gthhy4UxA9AcqaxaO0ZnLSRLM3PJhpT/NefX2XD/lb+4+rlLKkoGPXYgN/5U+Qsz/5UchaQWV3cXf2DFOZ6kh52tUioTEAp1TBikzNnaDoEp/VabDvSSVGuMfHPLpzWmKeUSro8ucVQGMqhwnvtk5D6sLCO/+mGQ1y9qop3nlk75rE52S78Xrejv0QbOk6tx8Ii4PfR1hsi4kDds5F09oUptikEBYkZiwYRORdQIuIRkX/DkN3QjIHTpMq3NXSyqqbItjsOGG7Ma3RIaKK7P0I4qmwJQzk9wT0VY2FJftSV5nL7W1ZM+LcSKPDSHHTG//VoNLT34XVnDXlBUyXg96JU4rpZ6aSjzz5dKEjMWNwC3IoxQ+IYhrT4rbataBow3GuRfs8iOBBmb3PQlv6KeMryvbiyxDGNeUOzt5PckAdxpbMO1QyaSmJ/eVUB5y8s4873rBnynMaj3O91dGPe8c4BqotyTvkGyRqv6uSQm0VnX9i2HgtIrBqqFXiPbSuYhjip1+Llo10oha3JbWCoWetElzM+VHY15AF43S6y3VmOnP8wGIkRHIhM2rMozsvmFzeenfDx5X4f28xGTydyonuAWQWnroFWXpA5jXld/WFOs6l7GxIwFqZc+EeBuvjjlVJX27aqDMdJvRZW57bdxgKMvMWJ7vQbSIiTvLAhDAXgd6jkh6ULZdf7trCqhJRStoY3p0pT9wBn1pWc8nUszyITjIWdirOQWAf37zHmUvwRcH6WxyE4Rar8pSOdzCvLs02JMp6KAh97HaJGaqdnAc4VE7RLPHEk5QVe+sOGfHciYatUopSiuTs05BWcCsOVX87Nz4DRV9Ifjtr6OU/EWAwope6wbQXTlNriHDbXt6d1DUoptjV0csHCspS8XkWhj7/tbXHE3WZrT4gsGVZUTTbGACTnhaGGpT7sleEY6rUIhhxnLDr6wgxGY0NzVk4Fn8dFgc/ZlV8A3WaZfqGNYahEEtz/KyJfNpvz1lgP21Y0Tagpzk17r8Wxzn5ae0K2yJKPRmWhj77BKEEHJH5be0KU5HmTOms8Hqd6FnaH3yyG7rgdmOS2yreTkbOAzBivanf3NiTmWZyGOfGO4TCUMn/WjIFVPnuso99Waz8eQ5PxUpCvgOEPZ1PXQNoHP02lMW0y+H1u2lrTn5MaiaUca3sYysHhmaZgco1Fud/nSKMYz7A8eXrDUO8A5mmJ8skRL1W+rGr0Tli7+cO2Y/h97jE7cZNNpdnF3dg1wMJZ6dWatEsXyiLf69Qw1OR1oaaCFYZyYnimacizSM7/f8DvZftR51Z+AXT22Tv4CBILQ70CpObWdBqR7iFIGw+18+TuZm65cD7Z7tTMXHDSLG67dKEsnByGKsmbnC7UVCjIcZPtznJk/0GT6QUka3xwwOwpUUol5Xp24JQwVBHwqohsAob+MnTp7PgU5XrIS1OvhVKK//zzbmYVePngeXNT9rpW9YkTxqvapQtlUeBz0zMYcZTKLhhhKLtDUGCUhztVZO9E9wCledlJu0kq9xuVX72DUcfO4h72LNIbhvqyba8+jTF6LXLT0mvx2M4mXjrSyX+99TRysu1RoBwNn8dFaV522uda9A1G6A9HbdGFsvD7PCgFPYORtOdn4pmK1MdUKfc7U/KjOUkNeRbxE/OcayzCuLOEPBs/74l0cP/Ntlef5hiNean1LCLRGP/92KvMD+TxdhvnV4yFEybmtQYnP/xnssRPy3OSsWjrCXFaTWqixuV+HwdaelLyWpPB6N5O3o3CcOXXAHPL8pJ23WTS2W805NlZsp7IPIugiHSbjwERiYpIt20rmkbUFOcMqV+mioc2H+VgSy+fvXwJ7jTMh3bCxDyrzLHMBl0oi2HlWWflLVIVhgLnDgZq6g4NqSAng6FkvoPLZ7v6wrZXXU74baKU8iulCpRSBUAO8DZADz9KgJriXIIp7LXoG4zwvSf3csacYi5ZNislrzmSikJf2sUErS7mgJ3VUEPKs86piApFogQHIikzFuV+L139YQbCzplYEI7GaOsNJS25Dc7uKbHo7B+0XaVhUreeyuD3GMOQNBOQaqnyezYcojkY4nNvWpK2DuqKAh8dfen9AklFY5oTZco7eg3DVWJzQ56FE0X2DL0qkupZFOV48LjE0Z5FZ1+YIps9i0SEBN8a92MWsBZwXlbLgcRLlS+vKrT1tdp7B/nR3w5yybJZrE2CgNpUsT6kTd0DzClNT3x3WB/J3moowFHKs229qdGFsoiX/DjV8aXJwqrES2bOIitLKMt3ZuWXRWdfmMUV9vY2JZLaf3Pc8whQD1xjy2qmGanstfjJ3w/SNxjhM5cttv21xiN+Yl66jEVT9wCFOR5b+0ucmLMY9qjs1YWyiK8ScgrN3cnt3rZwan7Goqs/bGv3NiRWDXWDrSuYxgz3Wtgfhtp6uIM1s4vT3jltTcxLZ6/FjqNdLK209/fgxAFIwyKCqQ5DOSfQYDWEJkNEMJ5yv5djnc55n/GEozF6QhFbR6rCOMZCRL40znlKKfU1G9YzrRjutbDfs6hv6+X8hQHbX2ciKuIkP9JBTyjCzuNd3HrxAltfJ8fjwpUljkpwp0oXyqI0z0uWOGuKXFMwhMclSVcbDvi9jh321JWC7m0YP8HdO8oD4EbgsxNdWERqReRpEdklIjtF5OPm9hIReUJE9pn/FpvbRUTuEJH9IrIjXtlWRD5gHr9PRD4wxfeaFlLRa9E3GKGpO0RdafrjxvleN/led9okP7Yd6SSmsD1vIyLkO2wAUltPCHcKdKEsXFlCab6zxqs2dQ1Q7vclvas+kO+lrXeQSNR5I32s7u3CdFVDKaW+bT2AuzDKZm8AHgTmJXDtCPAppdQyYB1wq4gsA24DnlJKLQSeMn8GeBOw0HzcDPwQDOOC0UV+NnAW8GXLwGQCtSX2d3EfaTeun64cwUiM8tn0GItN9e1kCaxJgSy70/Sh2nsHKU6BLlQ8TuviTnZDnkWgwIdSw6E+JzGsOJvGPgvTC/g6sAMjZLVGKfVZpVTzRBdWSjUqpbaaz4PAbqAaIzl+n3nYfcC15vNrgJ+b5bkvAEUiUolRpvuEUqpdKdUBPAFcPtk3mi5qinMIDkRs/SOrN6WyndJdWlHgS1vOYsvhDpZUFKRkIE+BzzN0V+cEmroHbNXDGo1yhyV+m7oHklo2a2H17DjpvVoMGYt0haFE5FvAJiAInKaU+or5ZT1pRKQOOB14EZillGo0d50ArO6xaqAh7rSj5raxtmcESysNefDdjfY1vde3GRHC2Q4IQ0H6PItINMbWIx2cWZcax3NuII8DLb0TH5giXj0RZPGs/JS+Zrnf56gv0Kbu5DbkWTixp8RiSHHW5mqo8TyLTwFVwBeA43GSH8HJyH2ISD7wMPCvSqnXnKcMzd+k6P6KyM0isllENre0tCTjkknBMha7jttnLA639VKal+0YjaLKQh/NwYGUx3d3NwbpG4ymrM9kaYWfI+19jqiIau8dpLFrIOWzU8oLvLT1hIjG0i/f3ROK0BOK2OxZOCfkZjGcs0iTZ6GUylJK5cTLfZgPvyn9MSEi4sEwFL9USv2fubnJDC9h/muFtI4BtXGn15jbxto+cr13KaXWKqXWBgLprwqyKMnLprLQx87jXba9Rn1rH3Mc4lWAUeMeU8a0ulSyyZx5vjZFnoU1VGrPiWBKXm88LM91WaW9zZ8jKfd7ianhhsB00mRDQ56F1VNS3+a86Yhd/WGyBPw2K+La1rUkht7E3cBupdR34natB6yKpg8Af4jb/n6zKmod0GWGqx4DLhWRYjOxfam5LWNYVlnALhvDUIfbeqlzSHIbhnstGlOsEbX5cDs1xTlDE/vsZonZy/HqifTralqeq939JSMJWF3cDqiIarKpIQ8M+f2LFwe497l66ludE3oEI2dRmOOxvbDBTlnS8zBnd4vINvNxBfBfwCUisg94o/kzwKPAQWA/8BPgwwBKqXbgaxj5k03AV81tGcPyqgIOtPTaopc0EI5yvGuAOockt2H4w5rKvIVSik31HZyZQqmT6qIc/F43rzam37PY1dhNRYEvZd3bFk7q4rbTWAD851tX4nEJn/rNdkeE3SwMeXL7e2ts81uUUhuAsUzdG0Y5XgG3jnGte4B7kre61LKsqoBoTLG3KcjKJM8aGC6bdU4YKh1d3Efa+2gJhlIWggKj12JJpd/W4oVE2XW8Oy2z3sv9zonlW+NU7TIWFYU+vnrNCv7119v4yd8PcsuF8215ncnS2TdoeyUU2OtZaEysOPJOG5LclkvspDBUSV422a6slHoWm+qNQr1UehZgFDC8eiKY1vnMA+Eo+1t6WFaZemPhJPnuE10DQ02hdnHN6iouX17Bdx7f64jwI1i6UNpYTAtqio1whR0VUYfNhJuTjIWIMKvQm1LPYnN9O4U5HhYEUls6uqSigJ5QJC2z1i32NfUQjam0eBY+j4vCHI8jymebg/Y05MUjItz+lhUU5Lj55K+3MxhJf0d3R5/9syxAG4uUkJUlLK2yJ8ld39ZLca7H9rK5yVJZkJNSfahN9e2cMac4pd3LEJ/kTl/eYlejUWmXDs8CnNPFfaIrubO3x6I038s33nIauxq7+X9/3Wf7601EZwqm5IE2FiljWWUBuxu7iSU5MVbf1usYmY94ZqWwMa+9d5ADLb0pzVdYLDZVfl9NY95i1/Fu8r1uZqdppkR5gTNmPTR1h5KuNjsWly6v4G1rarjzmQNsT6PAYCQaIzgQ0TmL6cSyqgL6BqND3dbJor61zxECgiOZU5LL8c7+ofkCdrLlcHryFQB5XjdzSnPT7Fl0s7TSn3KvysIJXdyxmKI5OEB5iowFwJevXkZZfjbfeWJvyl5zJN2mNpnOWUwjrBBBMkNRoUiU4139jvQs3rG2hphS3L3hkO2vtbm+nWxXFqdVp7YhzWJJRfoqomIxxe7GYNpCUDCsD5XOJH9H3yDhqKLC5pxFPAU+D1etrOL5g230D6ZnjLDVva1zFtOIhbPycWdJUpPcDe39KAV1ZQ70LErzuHJlFfe/cJiuPntnPmyqb2dlTSE+j8vW1xmLpZUFHGrrTcsXRkOHITeSjuS2RcDvZTASo7s/fbInJ2zusRiLixYHGIzEeOFgW0pf18LShUpFzlIbixThdbtYOMufVM/CiWWz8dxy4Tx6B6Pc/+Jh215jIBzl5WNdaZ07vqSiAKVgb1PqQ1FWOXaqZT7iCTig18Iq3Z1lgy7UeJw1t4Qcj4tn9kwoxG0LXSmSJwdtLFLKssqCpPZaWPkPpxqL5VWFXLgowD0bDtnSvQ6wvaGTcFSlTGl2NJamUfZj1/FuXFnCwhSrzcZjqbymM2+RLs/C63Zx7vxSnt7TkpYwXGe/DkNNS5ZVFdASDCXtDuxwWx8FPndKKiGmyocvmk9b7yC/2dww8cFTYLOZ3D5jTvqMRW1xLrnZLnanQfZjV2M3CwL5aQvBwbB8dzo9C0vqw+ooTyUXLQ5wpL2PQ2nQjLJmWdg9fxu0sUgpy4ZmWyTnS6W+rZe6sjwMzUZnctbcEtbMLuLHzx60RbJ8c307C8vzU3JnNRZZWcLiCn/aPIt05itg+As6neWzxuCnbDyu1H+lXbS4HIBn9qR+NEJnXxgRUjLsSxuLFGJ9qJOV5K53mNrsaIgI/3LRAo529PPIjsaJT5gkuxuDaauCimdJReplP9p6QpzoHkhrJRQYc9dzPK60Sn40dYdSHoKyqC3JZV4gj2f2pt5YdPWHKfB5cKWgbFobixRSmOOhpjgnKbMtBiMxjnX0O7LHYiRvWFLOwvJ8fvjMgaR+mXb1hTnRPcDiitTKco/Gsko/neZ6UoXloabbsxARygvSO141Vd3bY3HRonJeSEMJbUeKRARBG4uUk6zZFkc7+ogpHNljMZKsLOGWC+ezpynI00msGtnbbHxZLnKAsVhi3t2nUq7ckvlYmmbPAoxQ1LHO9OljGbpQaTQWaSqh7exLjYggaGORcpZVFXCotZe+wVOrSR8SEHRgj8VoXL26iuqiHH74zIGkXdPqmrYkN9KJ5d3sTmHeYtfxbioLfZTkpS9fY3H23FJeOtKRliT3YCRGa8+g7SKC45GuEtrO/jCFKcrXaWORYpZVGjX5I+UhXjrSwWXffZbN9YnNdbIqLzLBswDwuLK4+YJ5bKrv4G9Jiu3uPRHE73MPzc9IJwU+D9VFOSn2LLrTnq+wuHp1FTEFj9qQl5qIlh4j/JUqXajR8HlcnDO/NOV5i66+Qe1ZTFeWV58822JTfTvvu3sje5qC3Pd8Yg1sh9t68XvdlDrgrjJR3nVWLXNKc/n6I7uSUhm1pynI4ll+x1SDLa1MXUXUQDjKgZZelqc5X2GxaJafJRV+1m8/nvLXtgQr0xmGAiMUdbgttSW0xpQ8bSymJVWFPgpzPEMVUc8faOMD92yk3O/l8uUVPLmrKaEQVX1bH3PKch3zRZkIXreLz71pKfuae/jVplPru1DKmDzohHyFxdJKY3xuKGJ/knNvUzBtMyzG4urVVWw90kmDOb0xVTSnqSFvJBctskpoUxOKisVUygYfgTYWKUdEhpLcG/a1csO9G6kuyuHBf17H9efV0R+O8tTuif/YDjtUmnwiLls+i3XzSvjO43vo6p+6ZlRzMERnX9gR+QqLJRXG+Nx9TT22v9YuB8h8jOTNK6sAUu5dDHdvpy9nATC7NJd5ZXkp67cIDkRQKjXd26CNRVpYVlXAruNdfPC+TdSV5vGrm9dR7vdxZl0Jswq8PLJj/A9bOBqjIUPKZkciInzxqmV09odPaXDMHjPns8hJxiKFg5B2NXbj97qpKc6x/bUSpbYklzWzi/hjio1FU3cIj0sckei/cHGAFw622SZvE8+w1If2LKYty6sKCEcVC8vz+dVN6yjLN+6IXFnCladV8fSeFroHxr7rPtbRTzSmHN+QNxbLqwq57oxa7v1H/ZAY4mSxRPuc0GNhUVeah9edlZJBSDuPd7O0siBtMyzG4prV1bx6IphSUcWm7gHK/T5HhGQvWlxOKBLj+RSU0FpSH9pYTGOuOK2Sz1+xlAc+tI7iEXdDV62qZDAS44mdTWOePyQgWJaZxgLgU5ctItuVxTce3T2l8189ESTg9zribtLCZcp+JFMscjQGIzFeOdbFCgd0ro/kitMqyRJYvy013oVSioOtvVQ4oCIO4Oy5Jfg8WfwtBaGoIXnyHB2Gmrb4PC5uumDeqBr0p9cWUV2Uwx/HCUVZPRZzMjAMZVHu9/Hhixfw+K4m/nGgddLn7zUroZzGOfNKef5gG994dDfRJI/QtdjV2E0oEkvLGNmJCPi9nLegjPXbj6dE+uS7T+5je0Mnly6bZftrJYLP4+K8+WU8suO47XNchgcfac9iRiIivHlVFRv2tdLeOzjqMYdae8nNdhHIT29C71S58XVzqS7K4WuPTO6LNRYzK6EcaCz+7bLFvP+cOdz17EE+dN8mguOEE6fKFgco7Y7Hm1dVcaS9j202z6Z+4MUj3PHUPq5bW8PNF8yz9bUmwycuWURHX5iv/2mXra/TmcJZFqCNhSN586pKIjHFX145Mep+qxLKCTHaU8HncfG5K5awu7Gbz/x2R8K9Fw0dfQyEYyxxUL7CwuPK4qvXrOBr167g2X2tvPUH/+BwkueubzncTnVRTtpLRcfi8hUVZLuzbK2Kemp3E1/4/ctctDjA7W85zVGfhRXVhdx8wTx+s+Uof99nXzjKMhaF2ljMXJZVFjAvkDdqVckze5rZsL+V06qdU19/Klx5WiWfeOMiHt56lFvu35JQFYlVbeSkHouRvG/dHH7xwbNoDoa45s7nphRqGw2lFFsOdzgyBGVR4PNw8eIAj+xotCUU99KRDm59YCsrqgu585/WpEWWfCI+/oaFzCvL47aHX6Y3ZM+42c7+QfxeN+4UvX/n/ZY1RihqZRUvHGobajgCo9P7lvu3sLDcz+evXJbGFSYPEeHjb1zI165dwVOvNvO+u1+csP9ir2ksFpanbzpcIpy7oIw/3HoeZflerr9nE68cO3W14WOd/TR1hxwbgrK4elU1LcFQ0oX1DrX2cuN9m5lV4OOe688kz+tO6vWThc/j4ptvX8mxzn6+9dieU77es3tbTqoc7OoLp2T2toU2Fg7lzasqUQr+9LKhtfPKsS4++LNNVBXm8PMbz0qZ65kq3rduDt9/9+lsa+jknT9+/jVGciR7moLUluQ49osinrqyPB7653Mozc/mIw9sPeUchpWvWDPb2cbiDUvLyct2JbUqSinFTT/fjAD33XDWUMm5UzmzroT3nzOH+56vT1jzqg9X2wAAD6hJREFUbTSOtPXx/ns28obv/I3bHt4xpO6bSqkP0MbCsSwo97O0soA/bj/O/uYe3n/PRgpyPNz/obMd/yGZKletrOKe68/kSHsfb/vRPzjaMbpshFEJlTlhuJK8bO549+k0dPTz77975ZSqhLYc7iAv2+XIfE08Po+Ly1dU8rttx5KWu3j5WBf7m3u47U1LMqZs/DOXL6GqMIfPPrxjyo16j+00cpdvPb2a/9t6jIu/9QxfWb+Tox19FKdwQqRtxkJE7hGRZhF5JW7bV0TkmIhsMx9XxO37nIjsF5E9InJZ3PbLzW37ReQ2u9brRK5aWcnWI528+ycvkCXC/R86m6oi53Ts2sH5CwM8cNM6WoODfOfxvSftH4zEONjSy+IKZ4egRnJmXQmfvGQRf9x+nF+fgi7WlsMdrJ5dlLI49anw71csYVVNIR/71Ut8+/E9xE4xf/HEriZcWcIblzqjTDYR8r1uvvHW0zjQ0sv3p6hY8JedJ1hWWcC33rGKpz99EW9dU80vXjjM3qaelEYY7PyLuxe4fJTt31VKrTYfjwKIyDLgXcBy85wfiIhLRFzAncCbgGXAu81jZwSW1k4oHOUXN57F3Ay5mzpVVtcW8c4za1m//TiNXa8dqHOwtYdITDmybHYi/uXC+Zy/sIwvr985JFcyGXpDEXY3dnOGw0NQFqX5Xn75oXVct7aG7/91Px/+5dZTmuPy+M4mzqwrPqmR1elcuCjAW9dU8+O/HZz0vI/m4ABbj3Rw2fIKAKqLcvivt63kqU9eyPvWzeFtZ9TYseRRsc1YKKWeBRIN1F0DPKiUCimlDgH7gbPMx36l1EGl1CDwoHnsjGB2aS7fuW4VD958jiOmoaWSG183l5hS3Ptc/Wu2W1+yTpL5SJSsLOE7162mIMfDrQ9M/otzW0MnMQVn1JXYtMLkk+3O4ptvW8kXr1rG47tO8LYfPj+liXpH2vrY0xTkkmUVNqzSfj5y8QIiMcWvN07Oq3xiVxNKGeXI8dSV5fG1a1dw8eLyZC5zXNLhy35ERHaYYSrrFqkaiP8tHjW3jbV9xvDWNTWOkqFOFbUluVxxWiUPvHjkNUnhvU1B3FnCvLLMCkNZBPxe/vedqznQ0sOX/rBzUuduOdyBiOF5ZRIiwo2vm8s915/J0fY+rr3zuaHu40R5fJcRt3dKp/ZkmRfI5/yFZTyw8cikZrn85ZUT1JXmsmhW+v/eU20sfgjMB1YDjcC3k3VhEblZRDaLyOaWltROq9LYw80XzCMYirwmxr/nRJB5gTyy3c6P2Y/FuQvK+OjFC/jtlqND1U2JsOVwB4vK/RlbCXfR4nJ+edPZtARD3L3h0KTOfXxXE0sq/NSWZK7EzfvWzaGxa4AnExhBANDVH+b5A21ctrzCEU2HKf3EKaWalFJRpVQM+AlGmAngGFAbd2iNuW2s7aNd+y6l1Fql1NpAIJD8xWtSzsqaIs6eW8I9Gw4RNu/G9jhU5mOy3HzhfHyeLH730tGEjo/FFFuPdHCGg5vxEmFlTRFvWlHBvc/VJ6yd1N47yOb69oz1Kixev6ScqkIf97+Q2DTMp19tJhJTXLbCGaG3lBoLEamM+/EtgFUptR54l4h4RWQusBDYCGwCForIXBHJxkiCr0/lmjXp5eYL5nG8a4A/7WikNxShob3fkQKCkyXf6+aSZRX8aUcjg5GJwxL7mnsIDkQyJrk9Hh99/UKCoQg/+0di3sVTu5uIKTI2X2HhdmXxT2fPZsP+Vg60TDwg67GdJyj3e1ld44ywo52ls78CngcWi8hREbkR+G8ReVlEdgAXA58AUErtBB4CdgF/AW41PZAI8BHgMWA38JB5rGaGcPHichaU53PXswcdOcPiVHjL6VV09IV5du/EYVOniwdOhmVVBVy6bBb3bDg07twWiyd2NVFZ6GPFNJC4ue7MWjwumdC7GAhHeWZPC5cun+WYmSV2VkO9WylVqZTyKKVqlFJ3K6Xep5Q6TSm1Uil1tVKqMe7425VS85VSi5VSf47b/qhSapG573a71qtxJllZwk3nz2VXYzc/MyujpouxOH9hgJK8bH6/bdTI6mvYfLid0rzsjJalj+djb1hI90CE+0ZUu42kfzDKs/tauGTZLEfE7U+Vcr+Py1dU8tstR8ethnt2bwv94ehQyawTyNwsoWbGcM3qasryvazffhyfJ4va4unxhelxZXHVykqe2NU0oQzI1sMdnDGneFp8YYKhzPqGJeX8dMMhesYR2tuwv5WBcIxLMzwEFc/7z5lDcCAyrhTKYzubKPC5WTevNIUrGx9tLDSOx+dxcf25cwBj5rZT3PJkcM3qakKRGI+NMxmxtSdEfVvftAhBxfOxNyykqz/Mff+oH/OYJ3adwO9zc/a8zOktmYi1c4pZUuHn588fHlX6JRyN8dSrTbxx6SxHKeo6ZyUazTi8d90ccrNdLJ9mPSdrZhcxuySX3780dihq6zTKV8SzqraIixYH+OnfD44q4x2NKZ7a3czFi8sd9aV5qogI7103h12N3Ww9cvKAqI2H2unsC3Opg0JQoI2FJkMoys3mD7eex6cvW5LupSQVEeHa06t57kArTWMo7W453EG2K8uRM7dPlY+9YSEdfeFRE75bj3TQ1jvIpcszu2R2NK49vZp8r3vU9/3YzhP4PFlcuMhZLQDO13jWaEwWToOS2dG4dnUVdzy1jz9uP86Hzj95POiWwx2sqC7A53GlYXX2smZ2MecvLOOuZw+yqMLP6bVFFJlKqo/vPIHHJY770kwG+V43b11TzQMvHqElGKK2JJfZ5uPxnU1cuChATraz/r+1sdBo0sy8QD6ragr53UvHTjIWj+88wfajndxw3tw0rc5+PnXpYt591wvc8LNNAMwL5LFmdjHP7W/l3Pll+H2Z2bE+ER99/UJ6Q1H2t/Tw2M4TtPcOS6DctsJ5HrQ2FhqNA7hmdTVffWQX+5qCLJzlRynFD545wP88voeV1YXcNIrHMV1YXVvEpi+8kR1HO3npSCdbD3fw1O4mOvrCfPKSRelenm0E/F6+fd2qoZ+DA2Ea2vtp7x3k3PnOqYKy0MZCo3EAb15Vxe2P7ub3247x0dcv5LMP7+AP245z9aoq/vvtK6dlCCqefK+bc+eXce78MsCYitfSEyIwTQd9jYbf52FZlXO9KG0sNBoHEPB7ed2CMn639Rgb9rexvaGTT1+2mA9fNH/a9FZMBhGh3O9L9zI0cehqKI3GIVx7ehXHuwbY1xTkx+87g1svXjAjDYXGmWjPQqNxCG9aUcneph6uXlU144ZdaZyPNhYazf9v795CrKriOI5/fzimZZGWUZLWFIllN7UyLYkuFBbRk9CNsBCC8sEgqKSIgnoIIpMoJcoKkorsig+ZmQX1YI2peZm0m6BiTUEhBWnmv4e1Dh4vsWfmHNtnH38fOJy91z4z8/8z68z/7LX3rNUiBg8cwP1TW+8uGDPwMJSZmfWCi4WZmRVysTAzs0IuFmZmVsjFwszMCrlYmJlZIRcLMzMr5GJhZmaFdLBl/apO0i/AgauK9N5w4NcmhVO2dsoF2iufdsoFnE8r620up0bEQRcQacti0ShJXRFxYdlxNEM75QLtlU875QLOp5U1IxcPQ5mZWSEXCzMzK+RicXDPlx1AE7VTLtBe+bRTLuB8WlnDufiahZmZFfKZhZmZFXKxMDOzQi4WdSRNlbRR0neSHig7nr6StEBSj6R1dW3HSVoq6dv8PKzMGHtL0ihJyyVtkLRe0qzcXtV8Bkv6QtKanM+juf00SStyn3tD0hFlx9pbkgZIWiVpcd6vci6bJa2VtFpSV26rZF8DkDRU0iJJ30jqljS50XxcLDJJA4BngWuBscDNksaWG1WfvQxM3a/tAWBZRIwGluX9KtgN3BsRY4FJwMz8+6hqPjuBKyPifGAcMFXSJOAJYE5EnAH8BswoMca+mgV01+1XOReAKyJiXN3/I1S1rwHMBT6IiDOB80m/p8byiQg/0kX+ycCSuv3ZwOyy4+pHHp3Aurr9jcCIvD0C2Fh2jP3M6z3g6nbIBzgK+Aq4mPRftR25fZ8+2MoPYGT+g3MlsBhQVXPJ8W4Ghu/XVsm+BhwL/Ei+galZ+fjMYq+TgS11+1tzW9WdGBHb8/ZPwIllBtMfkjqB8cAKKpxPHrZZDfQAS4Hvgd8jYnd+SZX63NPAfcCevH881c0FIIAPJa2UdGduq2pfOw34BXgpDxO+IGkIDebjYnEYifSRolL3Sks6GngLuCcidtQfq1o+EfFPRIwjfSqfCJxZckj9Iul6oCciVpYdSxNNiYgJpGHomZIuqz9Ysb7WAUwA5kXEeOBP9hty6k8+LhZ7bQNG1e2PzG1V97OkEQD5uafkeHpN0kBSoVgYEW/n5srmUxMRvwPLSUM1QyV15ENV6XOXAjdI2gy8ThqKmks1cwEgIrbl5x7gHVIxr2pf2wpsjYgVeX8RqXg0lI+LxV5fAqPzHR1HADcB75ccUzO8D0zP29NJY/8tT5KAF4HuiHiq7lBV8zlB0tC8fSTp+ks3qWhMyy+rRD4RMTsiRkZEJ+l98nFE3EoFcwGQNETSMbVt4BpgHRXtaxHxE7BF0pjcdBWwgUbzKftiTCs9gOuATaSx5AfLjqcf8b8GbAf+Jn26mEEaS14GfAt8BBxXdpy9zGUK6TT5a2B1flxX4XzOA1blfNYBD+f204EvgO+AN4FBZcfax7wuBxZXOZcc95r8WF9771e1r+XYxwFdub+9CwxrNB9P92FmZoU8DGVmZoVcLMzMrJCLhZmZFXKxMDOzQi4WZmZWqKP4JWZWT1LtFkSAk4B/SNMrAEyMiF2lBGZ2CPnWWbMGSHoE+CMiniw7FrNDycNQZk0g6QJJn+aJ6JbUTavwiaQ5krryugIXSXo7rynwWH5NZ153YGF+zSJJR+VjV+XJ4Nbm9UoGlZmnHb5cLMwaJ+AZYFpEXAAsAB6vO74r0hoJ80lTLMwEzgFuz0NaAGOA5yLiLGAHcLekwaQ1Sm6MiHNJw8Z3/Q/5mB3AxcKscYNIf/yX5inIHyJNpFdTm2NsLbA+IrZHxE7gB/ZOXrklIj7P26+SpjsZA/wYEZty+yvAPrOhmv1ffIHbrHEiFYHJ/3F8Z37eU7dd26+9B/e/eOiLidZSfGZh1ridwAmSJkOaWl3S2X38HqfUvh64BfiMtLJZp6QzcvttwKfNCNisr1wszBq3hzQ19xOS1pBmyL2kj99jI2nRnW7SDKHzIuIv4A7gTUlr88+Z37ywzXrPt86alSwvG7s4Is4pORSz/+QzCzMzK+QzCzMzK+QzCzMzK+RiYWZmhVwszMyskIuFmZkVcrEwM7NC/wKfM2UaXj8fpAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7p1F7HdB2uJU"
      },
      "source": [
        "### Normalização"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "foeHRZr-2u-L"
      },
      "source": [
        "from sklearn.preprocessing import MinMaxScaler\n",
        "sc = MinMaxScaler(feature_range = (0, 1))\n",
        "training_set_scaled = sc.fit_transform(training_set)"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uQP8Q7MI21Y9"
      },
      "source": [
        "### Criar a estrutura de dados com janela 15 e output 1"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C6Ge9ql922cb"
      },
      "source": [
        "window = 15\n",
        "trainSize = len(training_set_scaled)\n",
        "X_train = []\n",
        "y_train = []\n",
        "for i in range(window, trainSize):\n",
        "    X_train.append(training_set_scaled[i-window:i, 0])\n",
        "    y_train.append(training_set_scaled[i, 0])\n",
        "X_train, y_train = np.array(X_train), np.array(y_train)"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LPBC2LoG24dU"
      },
      "source": [
        "### Reshaping"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oGwgQUDbdmnS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8a194dd1-7b52-4c60-c09b-f9428394c965"
      },
      "source": [
        "X_train.shape"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(45, 15)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b9ZfLlGH25fr"
      },
      "source": [
        "X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1))"
      ],
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RxU0axM9diHO",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "18f23cde-f5fb-449e-c469-39bafcfae784"
      },
      "source": [
        "X_train.shape"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(45, 15, 1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fZhjp7Tf282b"
      },
      "source": [
        "## Parte 2 - Construção da RNN"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kordpD0_3H7b"
      },
      "source": [
        "### Construção da Rede completa"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EskB17uo3Iso",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "781515fb-4dc2-4cfc-bd2b-78d4293805b7"
      },
      "source": [
        "# Inicializar a RNN\n",
        "regressor = Sequential()\n",
        "\n",
        "# Adicionar a primeira camada LSTM e Dropout \n",
        "regressor.add(LSTM(units = 48, return_sequences = True, input_shape = (X_train.shape[1], 1)))\n",
        "regressor.add(Dropout(0.2))\n",
        " \n",
        "# Adicionar a segunda camada LSTM e Dropout\n",
        "regressor.add(LSTM(units = 80, return_sequences = False))\n",
        "regressor.add(Dropout(0.2))\n",
        " \n",
        "# Adicionar a terceira camada LSTM e Dropout\n",
        "#regressor.add(LSTM(units = 50))\n",
        "#regressor.add(Dropout(0.2))\n",
        " \n",
        "# camada de saída\n",
        "regressor.add(Dense(units = 1))\n",
        " \n",
        "# Compilar a rede\n",
        "regressor.compile(optimizer = 'adam', loss = 'mean_squared_error')\n",
        "\n",
        "# Visualizar a rede\n",
        "regressor.summary()"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "lstm (LSTM)                  (None, 15, 48)            9600      \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 15, 48)            0         \n",
            "_________________________________________________________________\n",
            "lstm_1 (LSTM)                (None, 80)                41280     \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 80)                0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 1)                 81        \n",
            "=================================================================\n",
            "Total params: 50,961\n",
            "Trainable params: 50,961\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5j1w28g83iSR"
      },
      "source": [
        "### Treinamento"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xWKAuKyp3jXv",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f2203ba0-87ae-4884-b155-905010f65ee7"
      },
      "source": [
        "regressor.fit(X_train, y_train, epochs = 500, batch_size = 32)"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/500\n",
            "2/2 [==============================] - 3s 23ms/step - loss: 0.1271\n",
            "Epoch 2/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0744\n",
            "Epoch 3/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0547\n",
            "Epoch 4/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0689\n",
            "Epoch 5/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0697\n",
            "Epoch 6/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0617\n",
            "Epoch 7/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0591\n",
            "Epoch 8/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0644\n",
            "Epoch 9/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0633\n",
            "Epoch 10/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0666\n",
            "Epoch 11/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0605\n",
            "Epoch 12/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0618\n",
            "Epoch 13/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0520\n",
            "Epoch 14/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0503\n",
            "Epoch 15/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0536\n",
            "Epoch 16/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0570\n",
            "Epoch 17/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0524\n",
            "Epoch 18/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0454\n",
            "Epoch 19/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0457\n",
            "Epoch 20/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0474\n",
            "Epoch 21/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0374\n",
            "Epoch 22/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0380\n",
            "Epoch 23/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0349\n",
            "Epoch 24/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0374\n",
            "Epoch 25/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0265\n",
            "Epoch 26/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0278\n",
            "Epoch 27/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0180\n",
            "Epoch 28/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0187\n",
            "Epoch 29/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0309\n",
            "Epoch 30/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0225\n",
            "Epoch 31/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0267\n",
            "Epoch 32/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0231\n",
            "Epoch 33/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0190\n",
            "Epoch 34/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0241\n",
            "Epoch 35/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0147\n",
            "Epoch 36/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0199\n",
            "Epoch 37/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0170\n",
            "Epoch 38/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0169\n",
            "Epoch 39/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0158\n",
            "Epoch 40/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0187\n",
            "Epoch 41/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0169\n",
            "Epoch 42/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0153\n",
            "Epoch 43/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0192\n",
            "Epoch 44/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0196\n",
            "Epoch 45/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0153\n",
            "Epoch 46/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0188\n",
            "Epoch 47/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0116\n",
            "Epoch 48/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0126\n",
            "Epoch 49/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0165\n",
            "Epoch 50/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0142\n",
            "Epoch 51/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0147\n",
            "Epoch 52/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0161\n",
            "Epoch 53/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0166\n",
            "Epoch 54/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0156\n",
            "Epoch 55/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0140\n",
            "Epoch 56/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0152\n",
            "Epoch 57/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0157\n",
            "Epoch 58/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0132\n",
            "Epoch 59/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0163\n",
            "Epoch 60/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0152\n",
            "Epoch 61/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0134\n",
            "Epoch 62/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0179\n",
            "Epoch 63/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0173\n",
            "Epoch 64/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0134\n",
            "Epoch 65/500\n",
            "2/2 [==============================] - 0s 29ms/step - loss: 0.0152\n",
            "Epoch 66/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0138\n",
            "Epoch 67/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0156\n",
            "Epoch 68/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0140\n",
            "Epoch 69/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0178\n",
            "Epoch 70/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0159\n",
            "Epoch 71/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0136\n",
            "Epoch 72/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0136\n",
            "Epoch 73/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0126\n",
            "Epoch 74/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0137\n",
            "Epoch 75/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0124\n",
            "Epoch 76/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0183\n",
            "Epoch 77/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0168\n",
            "Epoch 78/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0123\n",
            "Epoch 79/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0160\n",
            "Epoch 80/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0154\n",
            "Epoch 81/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0132\n",
            "Epoch 82/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0127\n",
            "Epoch 83/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0148\n",
            "Epoch 84/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0122\n",
            "Epoch 85/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0133\n",
            "Epoch 86/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0193\n",
            "Epoch 87/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0153\n",
            "Epoch 88/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0163\n",
            "Epoch 89/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0138\n",
            "Epoch 90/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0143\n",
            "Epoch 91/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0132\n",
            "Epoch 92/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0120\n",
            "Epoch 93/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0150\n",
            "Epoch 94/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0131\n",
            "Epoch 95/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0128\n",
            "Epoch 96/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0140\n",
            "Epoch 97/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0138\n",
            "Epoch 98/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0133\n",
            "Epoch 99/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0144\n",
            "Epoch 100/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0154\n",
            "Epoch 101/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0123\n",
            "Epoch 102/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0120\n",
            "Epoch 103/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0145\n",
            "Epoch 104/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0141\n",
            "Epoch 105/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0144\n",
            "Epoch 106/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0118\n",
            "Epoch 107/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0137\n",
            "Epoch 108/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0137\n",
            "Epoch 109/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0126\n",
            "Epoch 110/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0157\n",
            "Epoch 111/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0170\n",
            "Epoch 112/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0137\n",
            "Epoch 113/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0106\n",
            "Epoch 114/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0121\n",
            "Epoch 115/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0091\n",
            "Epoch 116/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0104\n",
            "Epoch 117/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0150\n",
            "Epoch 118/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0145\n",
            "Epoch 119/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0136\n",
            "Epoch 120/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0148\n",
            "Epoch 121/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0142\n",
            "Epoch 122/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0142\n",
            "Epoch 123/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0127\n",
            "Epoch 124/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0140\n",
            "Epoch 125/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0129\n",
            "Epoch 126/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0145\n",
            "Epoch 127/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0127\n",
            "Epoch 128/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0127\n",
            "Epoch 129/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0101\n",
            "Epoch 130/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0116\n",
            "Epoch 131/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0146\n",
            "Epoch 132/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0107\n",
            "Epoch 133/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0140\n",
            "Epoch 134/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0129\n",
            "Epoch 135/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0147\n",
            "Epoch 136/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0132\n",
            "Epoch 137/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0140\n",
            "Epoch 138/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0108\n",
            "Epoch 139/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0125\n",
            "Epoch 140/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0123\n",
            "Epoch 141/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0138\n",
            "Epoch 142/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0098\n",
            "Epoch 143/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0151\n",
            "Epoch 144/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0115\n",
            "Epoch 145/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0092\n",
            "Epoch 146/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0138\n",
            "Epoch 147/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0132\n",
            "Epoch 148/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0136\n",
            "Epoch 149/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0140\n",
            "Epoch 150/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0105\n",
            "Epoch 151/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0129\n",
            "Epoch 152/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0152\n",
            "Epoch 153/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0169\n",
            "Epoch 154/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0131\n",
            "Epoch 155/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0127\n",
            "Epoch 156/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0114\n",
            "Epoch 157/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0136\n",
            "Epoch 158/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0146\n",
            "Epoch 159/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0132\n",
            "Epoch 160/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0110\n",
            "Epoch 161/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0155\n",
            "Epoch 162/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0147\n",
            "Epoch 163/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0126\n",
            "Epoch 164/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0136\n",
            "Epoch 165/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0120\n",
            "Epoch 166/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0121\n",
            "Epoch 167/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0131\n",
            "Epoch 168/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0085\n",
            "Epoch 169/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0119\n",
            "Epoch 170/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0148\n",
            "Epoch 171/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0116\n",
            "Epoch 172/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0107\n",
            "Epoch 173/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0110\n",
            "Epoch 174/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0117\n",
            "Epoch 175/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0121\n",
            "Epoch 176/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0118\n",
            "Epoch 177/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0111\n",
            "Epoch 178/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0091\n",
            "Epoch 179/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0097\n",
            "Epoch 180/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0121\n",
            "Epoch 181/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0096\n",
            "Epoch 182/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0101\n",
            "Epoch 183/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0108\n",
            "Epoch 184/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0101\n",
            "Epoch 185/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0129\n",
            "Epoch 186/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0105\n",
            "Epoch 187/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0102\n",
            "Epoch 188/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0111\n",
            "Epoch 189/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0128\n",
            "Epoch 190/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0106\n",
            "Epoch 191/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0137\n",
            "Epoch 192/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0107\n",
            "Epoch 193/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0116\n",
            "Epoch 194/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0115\n",
            "Epoch 195/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0121\n",
            "Epoch 196/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0142\n",
            "Epoch 197/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0142\n",
            "Epoch 198/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0142\n",
            "Epoch 199/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0117\n",
            "Epoch 200/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0145\n",
            "Epoch 201/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0102\n",
            "Epoch 202/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0122\n",
            "Epoch 203/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0132\n",
            "Epoch 204/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0109\n",
            "Epoch 205/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0114\n",
            "Epoch 206/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0128\n",
            "Epoch 207/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0092\n",
            "Epoch 208/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0074\n",
            "Epoch 209/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0088\n",
            "Epoch 210/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0119\n",
            "Epoch 211/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0153\n",
            "Epoch 212/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0118\n",
            "Epoch 213/500\n",
            "2/2 [==============================] - 0s 17ms/step - loss: 0.0110\n",
            "Epoch 214/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0104\n",
            "Epoch 215/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0130\n",
            "Epoch 216/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0093\n",
            "Epoch 217/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0064\n",
            "Epoch 218/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0096\n",
            "Epoch 219/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0166\n",
            "Epoch 220/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0132\n",
            "Epoch 221/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0106\n",
            "Epoch 222/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0126\n",
            "Epoch 223/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0112\n",
            "Epoch 224/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0126\n",
            "Epoch 225/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0090\n",
            "Epoch 226/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0124\n",
            "Epoch 227/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0121\n",
            "Epoch 228/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0094\n",
            "Epoch 229/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0084\n",
            "Epoch 230/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0099\n",
            "Epoch 231/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0143\n",
            "Epoch 232/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0123\n",
            "Epoch 233/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0098\n",
            "Epoch 234/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0100\n",
            "Epoch 235/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0112\n",
            "Epoch 236/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0110\n",
            "Epoch 237/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0109\n",
            "Epoch 238/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0091\n",
            "Epoch 239/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0085\n",
            "Epoch 240/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0089\n",
            "Epoch 241/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0111\n",
            "Epoch 242/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0108\n",
            "Epoch 243/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0113\n",
            "Epoch 244/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0098\n",
            "Epoch 245/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0119\n",
            "Epoch 246/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0140\n",
            "Epoch 247/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0094\n",
            "Epoch 248/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0091\n",
            "Epoch 249/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0116\n",
            "Epoch 250/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0095\n",
            "Epoch 251/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0087\n",
            "Epoch 252/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0107\n",
            "Epoch 253/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0098\n",
            "Epoch 254/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0087\n",
            "Epoch 255/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0087\n",
            "Epoch 256/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0108\n",
            "Epoch 257/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0092\n",
            "Epoch 258/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0092\n",
            "Epoch 259/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0132\n",
            "Epoch 260/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0096\n",
            "Epoch 261/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0119\n",
            "Epoch 262/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0102\n",
            "Epoch 263/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0121\n",
            "Epoch 264/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0134\n",
            "Epoch 265/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0121\n",
            "Epoch 266/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0098\n",
            "Epoch 267/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0135\n",
            "Epoch 268/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0113\n",
            "Epoch 269/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0098\n",
            "Epoch 270/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0093\n",
            "Epoch 271/500\n",
            "2/2 [==============================] - 0s 31ms/step - loss: 0.0118\n",
            "Epoch 272/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0102\n",
            "Epoch 273/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0106\n",
            "Epoch 274/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0107\n",
            "Epoch 275/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0125\n",
            "Epoch 276/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0117\n",
            "Epoch 277/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0089\n",
            "Epoch 278/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0117\n",
            "Epoch 279/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0083\n",
            "Epoch 280/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0083\n",
            "Epoch 281/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0117\n",
            "Epoch 282/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0093\n",
            "Epoch 283/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0115\n",
            "Epoch 284/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0104\n",
            "Epoch 285/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0094\n",
            "Epoch 286/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0140\n",
            "Epoch 287/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0080\n",
            "Epoch 288/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0092\n",
            "Epoch 289/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0112\n",
            "Epoch 290/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0102\n",
            "Epoch 291/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0108\n",
            "Epoch 292/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0098\n",
            "Epoch 293/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0106\n",
            "Epoch 294/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0091\n",
            "Epoch 295/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0083\n",
            "Epoch 296/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0102\n",
            "Epoch 297/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0102\n",
            "Epoch 298/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0118\n",
            "Epoch 299/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0084\n",
            "Epoch 300/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0117\n",
            "Epoch 301/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0093\n",
            "Epoch 302/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0082\n",
            "Epoch 303/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0099\n",
            "Epoch 304/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0110\n",
            "Epoch 305/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0127\n",
            "Epoch 306/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0088\n",
            "Epoch 307/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0109\n",
            "Epoch 308/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0096\n",
            "Epoch 309/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0105\n",
            "Epoch 310/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0093\n",
            "Epoch 311/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0074\n",
            "Epoch 312/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0116\n",
            "Epoch 313/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0083\n",
            "Epoch 314/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0111\n",
            "Epoch 315/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0085\n",
            "Epoch 316/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0132\n",
            "Epoch 317/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0115\n",
            "Epoch 318/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0115\n",
            "Epoch 319/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0103\n",
            "Epoch 320/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0081\n",
            "Epoch 321/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0092\n",
            "Epoch 322/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0087\n",
            "Epoch 323/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0104\n",
            "Epoch 324/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0090\n",
            "Epoch 325/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0113\n",
            "Epoch 326/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0095\n",
            "Epoch 327/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0097\n",
            "Epoch 328/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0092\n",
            "Epoch 329/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0087\n",
            "Epoch 330/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0090\n",
            "Epoch 331/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0095\n",
            "Epoch 332/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0096\n",
            "Epoch 333/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0097\n",
            "Epoch 334/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0115\n",
            "Epoch 335/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0110\n",
            "Epoch 336/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0103\n",
            "Epoch 337/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0106\n",
            "Epoch 338/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0078\n",
            "Epoch 339/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0102\n",
            "Epoch 340/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0095\n",
            "Epoch 341/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0114\n",
            "Epoch 342/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0086\n",
            "Epoch 343/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0076\n",
            "Epoch 344/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0095\n",
            "Epoch 345/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0082\n",
            "Epoch 346/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0100\n",
            "Epoch 347/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0114\n",
            "Epoch 348/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0083\n",
            "Epoch 349/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0097\n",
            "Epoch 350/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0091\n",
            "Epoch 351/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0095\n",
            "Epoch 352/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0096\n",
            "Epoch 353/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0075\n",
            "Epoch 354/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0082\n",
            "Epoch 355/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0091\n",
            "Epoch 356/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0095\n",
            "Epoch 357/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0103\n",
            "Epoch 358/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0120\n",
            "Epoch 359/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0079\n",
            "Epoch 360/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0078\n",
            "Epoch 361/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0098\n",
            "Epoch 362/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0108\n",
            "Epoch 363/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0088\n",
            "Epoch 364/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0104\n",
            "Epoch 365/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0095\n",
            "Epoch 366/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0080\n",
            "Epoch 367/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0089\n",
            "Epoch 368/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0114\n",
            "Epoch 369/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0093\n",
            "Epoch 370/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0092\n",
            "Epoch 371/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0095\n",
            "Epoch 372/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0130\n",
            "Epoch 373/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0093\n",
            "Epoch 374/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0105\n",
            "Epoch 375/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0113\n",
            "Epoch 376/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0103\n",
            "Epoch 377/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0102\n",
            "Epoch 378/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0087\n",
            "Epoch 379/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0101\n",
            "Epoch 380/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0082\n",
            "Epoch 381/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0095\n",
            "Epoch 382/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0117\n",
            "Epoch 383/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0082\n",
            "Epoch 384/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0084\n",
            "Epoch 385/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0108\n",
            "Epoch 386/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0106\n",
            "Epoch 387/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0122\n",
            "Epoch 388/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0099\n",
            "Epoch 389/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0079\n",
            "Epoch 390/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0107\n",
            "Epoch 391/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0122\n",
            "Epoch 392/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0096\n",
            "Epoch 393/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0107\n",
            "Epoch 394/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0114\n",
            "Epoch 395/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0077\n",
            "Epoch 396/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0084\n",
            "Epoch 397/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0084\n",
            "Epoch 398/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0076\n",
            "Epoch 399/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0087\n",
            "Epoch 400/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0088\n",
            "Epoch 401/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0088\n",
            "Epoch 402/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0082\n",
            "Epoch 403/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0091\n",
            "Epoch 404/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0079\n",
            "Epoch 405/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0095\n",
            "Epoch 406/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0085\n",
            "Epoch 407/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0103\n",
            "Epoch 408/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0072\n",
            "Epoch 409/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0093\n",
            "Epoch 410/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0102\n",
            "Epoch 411/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0084\n",
            "Epoch 412/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0089\n",
            "Epoch 413/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0096\n",
            "Epoch 414/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0086\n",
            "Epoch 415/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0095\n",
            "Epoch 416/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0094\n",
            "Epoch 417/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0100\n",
            "Epoch 418/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0117\n",
            "Epoch 419/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0119\n",
            "Epoch 420/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0110\n",
            "Epoch 421/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0078\n",
            "Epoch 422/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0089\n",
            "Epoch 423/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0075\n",
            "Epoch 424/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0108\n",
            "Epoch 425/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0100\n",
            "Epoch 426/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0097\n",
            "Epoch 427/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0089\n",
            "Epoch 428/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0095\n",
            "Epoch 429/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0113\n",
            "Epoch 430/500\n",
            "2/2 [==============================] - 0s 18ms/step - loss: 0.0102\n",
            "Epoch 431/500\n",
            "2/2 [==============================] - 0s 27ms/step - loss: 0.0108\n",
            "Epoch 432/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0085\n",
            "Epoch 433/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0108\n",
            "Epoch 434/500\n",
            "2/2 [==============================] - 0s 27ms/step - loss: 0.0128\n",
            "Epoch 435/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0090\n",
            "Epoch 436/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0102\n",
            "Epoch 437/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0081\n",
            "Epoch 438/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0102\n",
            "Epoch 439/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0090\n",
            "Epoch 440/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0115\n",
            "Epoch 441/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0107\n",
            "Epoch 442/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0105\n",
            "Epoch 443/500\n",
            "2/2 [==============================] - 0s 34ms/step - loss: 0.0083\n",
            "Epoch 444/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0106\n",
            "Epoch 445/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0103\n",
            "Epoch 446/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0103\n",
            "Epoch 447/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0099\n",
            "Epoch 448/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0094\n",
            "Epoch 449/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0097\n",
            "Epoch 450/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0086\n",
            "Epoch 451/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0096\n",
            "Epoch 452/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0118\n",
            "Epoch 453/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0082\n",
            "Epoch 454/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0095\n",
            "Epoch 455/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0099\n",
            "Epoch 456/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0114\n",
            "Epoch 457/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0058\n",
            "Epoch 458/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0073\n",
            "Epoch 459/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0091\n",
            "Epoch 460/500\n",
            "2/2 [==============================] - 0s 29ms/step - loss: 0.0103\n",
            "Epoch 461/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0082\n",
            "Epoch 462/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0096\n",
            "Epoch 463/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0085\n",
            "Epoch 464/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0094\n",
            "Epoch 465/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0124\n",
            "Epoch 466/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0069\n",
            "Epoch 467/500\n",
            "2/2 [==============================] - 0s 28ms/step - loss: 0.0083\n",
            "Epoch 468/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0109\n",
            "Epoch 469/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0092\n",
            "Epoch 470/500\n",
            "2/2 [==============================] - 0s 24ms/step - loss: 0.0068\n",
            "Epoch 471/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0089\n",
            "Epoch 472/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0093\n",
            "Epoch 473/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0109\n",
            "Epoch 474/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0106\n",
            "Epoch 475/500\n",
            "2/2 [==============================] - 0s 26ms/step - loss: 0.0106\n",
            "Epoch 476/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0106\n",
            "Epoch 477/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0115\n",
            "Epoch 478/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0114\n",
            "Epoch 479/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0112\n",
            "Epoch 480/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0107\n",
            "Epoch 481/500\n",
            "2/2 [==============================] - 0s 19ms/step - loss: 0.0105\n",
            "Epoch 482/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0101\n",
            "Epoch 483/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0080\n",
            "Epoch 484/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0119\n",
            "Epoch 485/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0109\n",
            "Epoch 486/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0069\n",
            "Epoch 487/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0109\n",
            "Epoch 488/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0093\n",
            "Epoch 489/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0099\n",
            "Epoch 490/500\n",
            "2/2 [==============================] - 0s 37ms/step - loss: 0.0089\n",
            "Epoch 491/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0098\n",
            "Epoch 492/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0092\n",
            "Epoch 493/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0084\n",
            "Epoch 494/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0105\n",
            "Epoch 495/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0108\n",
            "Epoch 496/500\n",
            "2/2 [==============================] - 0s 21ms/step - loss: 0.0085\n",
            "Epoch 497/500\n",
            "2/2 [==============================] - 0s 23ms/step - loss: 0.0103\n",
            "Epoch 498/500\n",
            "2/2 [==============================] - 0s 25ms/step - loss: 0.0106\n",
            "Epoch 499/500\n",
            "2/2 [==============================] - 0s 20ms/step - loss: 0.0091\n",
            "Epoch 500/500\n",
            "2/2 [==============================] - 0s 22ms/step - loss: 0.0091\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f6e7f992b10>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gwpyImwA3yon"
      },
      "source": [
        "## Parte 3 - Fazer as previsões e analisar os resultados"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wkXRvDjT4V-B"
      },
      "source": [
        "### Base de teste"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bn_n72dh30K-"
      },
      "source": [
        "dataset_test = pd.read_csv('test.csv')\n",
        "test_pul = dataset_test.iloc[:, 1:2].values"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EOo84exH4bfs"
      },
      "source": [
        "### Construir a estrutura para teste"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ob_-mGA_B5px"
      },
      "source": [
        "dataset_total = pd.concat((dataset_train.iloc[:,1], dataset_test.iloc[:,1]), axis = 0)\n",
        "testLength = len(dataset_test)\n",
        "inputs = dataset_total[len(dataset_total) - testLength - window:].values\n",
        "inputs = inputs.reshape(inputs.shape[0],1)\n",
        "\n",
        "inputs = sc.transform(inputs)\n",
        "X_test = []\n",
        "for i in range(window, inputs.shape[0]): \n",
        "    X_test.append(inputs[i-window:i, 0])\n",
        "X_test = np.array(X_test)\n",
        "\n",
        "X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rtkB4NA74e1B"
      },
      "source": [
        "### Fazer as previsões"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nKIq1h-E4g9Z"
      },
      "source": [
        "predicted = regressor.predict(X_test)\n",
        "predicted = sc.inverse_transform(predicted)"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "guvk7yXr4o0g"
      },
      "source": [
        "### Visualizar os resultados de treino e teste"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tstXl2oS4rIo",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "c4cd13ba-25db-4ae3-846a-be18f42b01b3"
      },
      "source": [
        "allTargetData = np.vstack((training_set, test_pul))\n",
        "training_predicted = regressor.predict(X_train)\n",
        "training_predicted = sc.inverse_transform(training_predicted)\n",
        "allForecastedData = np.vstack((training_set[0:window], training_predicted, predicted))\n",
        "plt.plot(allTargetData, color = 'red', label = 'Real')\n",
        "plt.plot(allForecastedData, color = 'blue', label = 'Previsto')\n",
        "plt.xlabel(\"Tempo\")\n",
        "plt.ylabel(\"Numero de Mortes\")\n",
        "plt.title(\"Doenças Pulmonares\")\n",
        "plt.legend()\n",
        "plt.savefig('predictions_training_test.svg')\n",
        "plt.show()"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ytEVhQht4we3"
      },
      "source": [
        "## Parte 4 - Métricas de avaliação"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rqOonVj64ydu",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "853f85e3-723f-431e-9dc1-9517b97a8d50"
      },
      "source": [
        "rmse = math.sqrt(mean_squared_error(test_pul, predicted))\n",
        "print('RMSE: ', rmse)"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "RMSE:  168.13425899287557\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ODeley1K42WW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "986064a9-5f94-42c7-c747-1e092b8c9604"
      },
      "source": [
        "mse = mean_squared_error(test_pul, predicted)\n",
        "print('MSE: ',mse)"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MSE:  28269.12904708336\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L7_uU_8T44QA",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5a1711ac-9320-4fb7-c0b0-cfa52a8aee0a"
      },
      "source": [
        "mape = np.mean(np.abs((test_pul - predicted) / test_pul)) * 100\n",
        "print('MAPE: ',mape, '%')"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MAPE:  5.4638072521610574 %\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mgmtdSOq7rG0",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "aa32beda-0a33-48ec-8e6a-11d522d98162"
      },
      "source": [
        "mae = mean_absolute_error(test_pul, predicted)\n",
        "print('MAE: ',mae)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "MAE:  114.87921142578125\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}