TestEstadisticos.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd;\n",
    "import pingouin as pg;\n",
    "import seaborn as sns;\n",
    "import scipy.stats as stats;\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import scikit_posthocs as scp\n",
    "import statsmodels as st"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Lectura de los datos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "cnn_no_ll24_cv_10_3 = pd.read_csv('Resultados/lag24/NO/CNN/CNN-NO-LL24-Crosvalidacion-media24steps.csv',\n",
    "                                 index_col=0)\n",
    "gru_no_ll24_cv_10_3 = pd.read_csv('Resultados/lag24/NO/GRU/GRU-NO-LL24-Crosvalidacion-media24steps.csv',\n",
    "                                 index_col=0)\n",
    "lstm_no_ll24_cv_10_3  = pd.read_csv('Resultados/lag24/NO/LSTM/LSTM-NO-LL24-Crosvalidacion-media24steps.csv',\n",
    "                                 index_col=0)\n",
    "rsme_no_ll24 = pd.DataFrame({\"gru\":gru_no_ll24_cv_10_3['RMSE'],\n",
    "                             \"cnn\":cnn_no_ll24_cv_10_3['RMSE'],\n",
    "                             \"lstm\":lstm_no_ll24_cv_10_3['RMSE']})\n",
    "mae_no_ll24 = pd.DataFrame({\"gru\":gru_no_ll24_cv_10_3['MAE'],\n",
    "                             \"cnn\":cnn_no_ll24_cv_10_3['MAE'],\n",
    "                             \"lstm\":lstm_no_ll24_cv_10_3['MAE']})\n",
    "cc_no_ll24 = pd.DataFrame({\"gru\":gru_no_ll24_cv_10_3['CC'],\n",
    "                             \"cnn\":cnn_no_ll24_cv_10_3['CC'],\n",
    "                             \"lstm\":lstm_no_ll24_cv_10_3['CC']})\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## RSME"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1c1e8fd910>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rsme_no_ll24.boxplot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test estadísticos\n",
    "\n",
    "Primero hay que comprobar si las muestras proceden de una distribución normal. Para ello, aplicamos el test de Shapiro-Wilks, para determinar si las muestras proceden de una distribución normal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gru     (0.8127288818359375, 0.00011284616630291566)\n",
       "cnn       (0.808948278427124, 9.585390944266692e-05)\n",
       "lstm    (0.8350979089736938, 0.00030700559727847576)\n",
       "dtype: object"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rsme_no_ll24.apply(stats.shapiro)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Como se puede apreciar, todos los p-values (segundos valores en cada valor devuelto) es menor que 0.05 con lo que con una confinaza del 95% podemos rechazar la hipótesis de que las muestras proceden de una distribución normal. Esto también se puede ver por las gráficas de densidad, en las que se ve claramente que no siguen una distribución normal. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "dataSets = (rsme_no_ll24,mae_no_ll24,cc_no_ll24)\n",
    "fig, axes = plt.subplots(3, 1)\n",
    "for  df,i in zip(dataSets,range(3)):\n",
    "     df.plot(kind='density', ax=axes[i]);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "TODO: descrubir la forma de poner los títulos en cada figura. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Al no poder aplicar una técnica paramétrica, se procede a comprobar si hay diferencias significativas entre los tres modelos con el test de Friedman. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "FriedmanchisquareResult(statistic=6.199999999999989, pvalue=0.04504920239355807)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stats.friedmanchisquare(rsme_no_ll24.gru,rsme_no_ll24.lstm,rsme_no_ll24.cnn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Al ser el pvalue menor que 0.05, con un 95% podemos afirmar que existen diferencias significativas, vamos a proceder a localizar donde están esas diferencias significativas aplicando el test post-hoc de Nemenyi."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>gru</th>\n",
       "      <th>cnn</th>\n",
       "      <th>lstm</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>gru</th>\n",
       "      <td>-1.000000</td>\n",
       "      <td>0.128475</td>\n",
       "      <td>0.052581</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cnn</th>\n",
       "      <td>0.128475</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>0.900000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lstm</th>\n",
       "      <td>0.052581</td>\n",
       "      <td>0.900000</td>\n",
       "      <td>-1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           gru       cnn      lstm\n",
       "gru  -1.000000  0.128475  0.052581\n",
       "cnn   0.128475 -1.000000  0.900000\n",
       "lstm  0.052581  0.900000 -1.000000"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scp.posthoc_nemenyi_friedman(rsme_no_ll24, melted=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Haciendo los test por pares para localizar donde están diferencias significativas estas no se encuntran. Esto puede ocurrir al estar el resultado del test de Friedman cerca de límite."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Vamos a intentar una transformación logarítmica de lsos datos para ver si se cumplen los criterios de parametricidad. Empezaremos por comporbar la normalida"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gru     (0.8686501979827881, 0.0015635015442967415)\n",
       "cnn     (0.8588370084762573, 0.0009547794470563531)\n",
       "lstm     (0.8955364227294922, 0.006543213501572609)\n",
       "dtype: object"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.log(rsme_no_ll24).apply(stats.shapiro)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Como podemos ver una transformación logarítmica tampoco funciona. Ya sólo nos queda por comprobar una comparación por pares utilizando el test te wilcoxon."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>wins</th>\n",
       "      <th>ties</th>\n",
       "      <th>losses</th>\n",
       "      <th>diff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>gru</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lstm</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cnn</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      wins  ties  losses  diff\n",
       "gru    1.0   1.0     0.0   1.0\n",
       "lstm   0.0   2.0     0.0   0.0\n",
       "cnn    0.0   1.0     1.0  -1.0"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cols = rsme_no_ll24.columns\n",
    "pvalues = np.array([])\n",
    "rmse_ranks = pd.DataFrame({\"wins\":np.zeros(3),\n",
    "                     \"ties\":np.zeros(3),\n",
    "                     \"losses\":np.zeros(3)})\n",
    "rmse_ranks.index= cols\n",
    "rmse_ranks.index= cols\n",
    "for index1 in range(len(cols)):\n",
    "    for index2 in range(index1+1,len(cols)):\n",
    "        T,p = stats.wilcoxon(rsme_no_ll24.iloc[:,index1],rsme_no_ll24.iloc[:,index2])\n",
    "        test,padjust,a1,a2 = st.stats.multitest.multipletests(p,alpha = 0.05,method = \"bonferroni\")\n",
    "        if not test:\n",
    "            rmse_ranks.iloc[index1,1] =rmse_ranks.iloc[index1,1]+1\n",
    "            rmse_ranks.iloc[index2,1] = rmse_ranks.iloc[index2,1]+1\n",
    "        else:\n",
    "            if rsme_no_ll24.iloc[:,index1].mean() > rsme_no_ll24.iloc[:,index2].mean():\n",
    "                rmse_ranks.iloc[index1,0] =rmse_ranks.iloc[index1,0]+1\n",
    "                rmse_ranks.iloc[index2,2] = rmse_ranks.iloc[index2,2]+1\n",
    "            else:\n",
    "                rmse_ranks.iloc[index2,0] = rmse_ranks.iloc[index2,0]+1\n",
    "                rmse_ranks.iloc[index1,1] = rmse_ranks.iloc[index1,2]+1\n",
    "                \n",
    "rmse_ranks[\"diff\"] = rmse_ranks.wins-rmse_ranks.losses\n",
    "rmse_ranks = rmse_ranks.sort_values(by=\"diff\",ascending = False)\n",
    "rmse_ranks"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1c1ec59410>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mae_no_ll24.boxplot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tests estadísticos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "mae_no_ll24.boxplot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Comprobación de la normalidad"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gru       (0.7871785163879395, 3.8616872188868e-05)\n",
       "cnn     (0.7480366230010986, 8.466297913400922e-06)\n",
       "lstm    (0.8196220397949219, 0.0001526105625089258)\n",
       "dtype: object"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mae_no_ll24.apply(stats.shapiro)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Como podemos apreciar, al 95% de confianza podemos rechazar la hipótesis nula de que las distintas muestras proceden de una distribución normal. También lo podemos comprobar gráficamente. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1c1edc8810>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mae_no_ll24.plot(kind='density')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1c1eec8110>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYQAAAD7CAYAAABqvuNzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzdd3hUVf7H8ff0yaT3SkJCGpCEEggQei8CSrGBwO4qru4qLuu6uoplXduu+sOC7gq6VnCxUEXpIEioAQKEkJAC6b0nM8lk5v7+iAZjAgTJZFLO63nmeZzb5jOYme/cc+49RyZJkoQgCILQ48mtHUAQBEHoHERBEARBEABREARBEIQfiYIgCIIgAKIgCIIgCD9SWjvAr2EwGDh37hzu7u4oFAprxxEEQegSTCYTRUVFREREoNVqW6zvkgXh3LlzLFy40NoxBEEQuqS1a9cyZMiQFsu7ZEFwd3cHGt+Ul5eXldMIgiB0Dfn5+SxcuLDpO/SXumRB+KmZyMvLCz8/PyunEQRB6Fqu1tQuOpUFQRAEQBQEQRAE4UddsslIEATh1zCbzWRnZ1NTU2PtKBalUqnw8PDAwcHhhvYTBUEQhB6juLgYmUxGWFgYcnn3bCCRJAm9Xk9OTg7ADRWF7vkvIgiC0Iry8nI8PT27bTEAkMlk6HQ6fH19KSwsvKF9u++/iiAIwi+YTCZUKpW1Y3QIGxsbjEbjDe0jmoyELs9Ylk/p3s+oTT8NkoRN70hcxi9A7e5v7WhCJySTyawdoUP8mvcpzhCELq02I4HsNY9Sm5GAXf/R2EeNw5CdRPYHj1Gd+IO14wlClyLOEIQuqy4vjYIvXkHl4oXXnStQOrgC4Dz6Dgq+fpXCzW8iU2mwDR1q5aSC0DWIgiB0SeZ6PQUb/w+5jT3eC55DYevYtE5h64jXXSvI/fRpira8hfreV1E5iyFOhM7n9ddfZ8eOHTg7O+Pu7s6ECRNYvXo1zs7OaLVaZs2axbFjx3jllVcAWLRoEQ899BDDhg2zSB5REIQuqfzQBhrK8vG+5/lmxeAncrUWz3mPkb3mzxRvX43XXU/3mLZjoW32nshk17FMixx7cow/E4Zcuw9r7969xMfH880336DX65kzZw4TJkwgIyOD999/Hz8/PzZs2GCRfFcj+hCELsdYXkDF0a3YRYzBJqD/VbdTOXngMm4B+vQEas4f6sCEgnB9cXFxTJ8+HbVajaOjI5MmTQLA1dXVamO0iTMEocspO/gVyGS4jL/nuts6RE+lKmEvpfvXYhs+DJmiZ1xyKFzfhCHX/xVvSXK5HLPZ3GL5z+cpkMlkSJLU9PxGLyO94UwWPbogtLOGyhKqzx3AfuDEpk7ka5HJFbiMW0BDeSFVCfs6IKEgtE1sbCw7d+6kvr6e6upq9u/fT25ubrNtnJ2dSUtLQ5IksrKySE5OtmgmcYYgdCkVx7eBZMZx2Kw272PTZxAavzDKDn2N/YAJyBTiz16wvnHjxnHq1CnmzJmDo6MjHh4eaDSaZtvExsby9ddfM23aNAIDA4mOjrZoJvHJELoMqcFIVcIebMOHoXLybPN+MpkMpxFzKPjyFWqSj2LXb6QFUwpC25w6dYrevXuzbds2jEYjd955J6NHj2bp0qVN26jVat5+++0OyySajIQuo+biccz6auwHTrrhfXUh0ahcvKk4utUCyQThxgUGBvLNN98we/Zs5s6dyy233EJ4eLhVM4kzBKHLqDq9B4WDGza9I294X5lMjsPQWyjZ8T6G7GS0fmEWSCgIbefk5MQHH3xg7RjNiDMEoUtoqCxGn56AfdQ4ZPLWp/+7HvuoccjUWipP7W7ndILQPYiCIHQJ1UmHAQn7yHG/+hhytQ12/UZRkxSHuU7fbtkEobsQBUHoEmouHEHt0RuVi/dNHcd+wAQko4HqJHGjmiD8ksULQnV1NTNnziQ7O7vFuqSkJObOncvUqVN56qmnaGhosHQcoQtqqCqjLjsZ2/CbH79F4xuKytWXqoS97ZBMELoXixaEhIQE7r77bi5dutTq+scee4xnnnmGHTt2IEkSX3zxhSXjCF1UbcpRQMI2fPhNH0smk2E/YAJ12cnUl+RefwdB6EEsWhC++OILnn32WTw8PFqsy8nJwWAwMHDgQADmzp3L9u3bLRlH6KJqLhxB5eKDyq1XuxzPrv/oxuOK8Y0EoRmLXnb64osvXnVdYWEh7u7uTc/d3d0pKChosV1lZSWVlZXNluXn57dfSKFTMxlq0F9OxGnEre02WqnSwRVtr75UJx3CefTt7XJMQegOrHYfgtlsbvYBlySp1Q/8xx9/zKpVqzoymtCJ6C+dAcmMLrh9b9m37RtLyc4PqC/KFFNt9lBVZ/ZbrC/JfsAE7KPGXXMbSZJ47bXX2L17NwqFgjvvvJPdu3cTGRlJfHw8paWlrFixgrFjx/LEE09gZ2dHYmIiBQUF/PGPf2TevHntnttqBcHLy4uioqKm58XFxa02LS1ZsoQ5c+Y0W5afn8/ChQstnlGwPn3aaWQaHRqfkHY9rm3fEZTs+pDq84dwGSsKgtDxtm/fzsmTJ9m6dStGo5EFCxZQV1eH0Whk/fr17N27lzfffJOxY8cCjd9769atIyUlhcWLF3evguDr64tGoyE+Pp7o6Gg2b97MmDFjWmzn4OCAg4ODFRIK1iZJErXpp7HpHXndAenKq+pIzS7HZDLT28cRTxfdNbdX2jmjDehPzfk4nMfcJSbP6YHso8Zd91e8JR0/frxpPgS1Ws3mzZtZtGgRo0c39nGFhIRQXl7etP3IkSORyWSEhoY2W96eOrwgLF26lGXLlhEZGclrr73GihUrqK6upn///ixevLij4widmLEkB1NlMbqRV/8llF9Sw3+3JnLkXB4/GzaeiD6u3DsrguBeTlfd165vLMXfvUd94WU0nr3bMbkgXJ9SqWz2QyQ7O5va2tqmEU9/+SPlasvbNZPFjvwze/deaadbs2ZN03+Hh4fz1VdfdUQEoQuqTTsFgE2fga2uj79QwKufnsAsScyfEEJ0uCdKhYyzaSVsOZDGo28d4L7ZEcwaHdTq/rrQGPhuNbUpx0RBEDrc0KFD+eSTT7jrrrtoaGjgvvvuo7q62qqZxOB2QqelT09A5eqLyrFl31JCShEv/PcYvTztWPHbYXj8rIkoLMCFacMDeON/p1i96SyG+gZunxja4hhKOyc0fqHUpBzHefQdFn0vgvBLkydP5ty5c8ydOxez2czixYv57rvvrJpJFAShU5JMDRiyklpt480urOLFj47i627Liw+OxF6nbrGNnU7N334Twxufn+STb5NwcdAycWjLzmPb0BhK935KQ0URSkf3FusFwZKWL1/O8uXLm54vWLCg6b/9/PyaWldeeeWVZvtZauY0MZaR0CnV5acjGQ1oAyKaLTc2mHj103iUCgXPLR3RajH4iUIuY9mdgxgQ4sY7XyWQkVvRYhtdaAwANSnH2/cNCEIXJAqC0CkZLicCYOPfr9ny9btSSM+t4E93DcLNyea6x1Ep5fxl4RDsbFS8+tkJ6oymZuvVrj6oXH2pTTnWfuEFoYsSBUHolPSZiajc/FDYOjYtyy2q5ut9qYyL9iOmv1ebj+Vkr+FPdw8mq6Car/ZcbLHeNiwG/eVETHrrdugJgrWJgiB0OpLZhCErCRv//s2Wv7/lHGqVnN/N7H+VPa9ucJgHYwf58dXei2QXVjVbpwuNAclMbdrJm8otCF2dKAhCp1OXn4FUb0AbcOWL/8LlUo6fL2D+hBCcHbS/6rj3zu6PRiVnzaZzzZZrfIJR2DqJZiOhxxMFQeh0DJmN/Qfan/UffL4jGQdbNTNHtX5PQVs4O2i5Y1IYJ5MLOZta3LRcJpOjC45Gn56AZBJzcgg9lygIQqdjuJyIytUHpZ0zAEkZpZxMLmTuuGBsNDd3pfQtowJxddTyybfnkX52a7MueDDmuloMOZa5nE8QWnP06FEWLVrU6rqsrCyefPLJDs0jCoLQqUiShCEnGW2vK2cHmw6kYmejYsbIwJs+vkal4K7JYVy4XMbx81eGW7cJjAK5ktpU0Y8gdA65ublkZWV16GuKG9OETsVYmotZX43Gt/HO4sKyWo6cy2fO2D43fXbwk0kx/mzYn8r/diUztJ8nMpkMuUaH1r8vtakncZ3Q+i82QbCUDz/8kI0bNyKXy4mKiuL555/nhRdeIDs7m7///e9MmzaN//znP6hUKrKzs5kwYQI6nY7du3cDsHr1atzc3G46hygIQqdSl93YZKP9sSB8F3cJJIkZsTd/dvATpULO3HHBvPNVAufSSogMbvwg6foMpnTPx+Ku5R7i+4wj7MuIs8ixxwfGMjawbVO+mkwm3nvvPQ4ePIhCoeCpp56ioKCAFStWsGrVKp599lmOHj1KQkIC27Ztw8nJidjYWB5//HE2bNjA3/72N7Zt28aSJUtuOrdoMhI6FUNOCnKtLSo3X+qMJnYcucSwCO9mYxW1hwlDeuFkp+GrfVfuS9AFDwauDKonCB1BoVAwaNAg5s+fz6pVq/jtb3+Lp6dni+1CQ0Px9vbGxsYGZ2dnRowYAYCPj0+LWSV/LXGGIHQqdTnJaHxCkMnkHErIoqrWyMxR7Xd28BO1SsGs0UF8+l0SGbkVBPo4onL1RenoQW1qPA6Dp7T7awqdy9jA4W3+FW9p7777LqdPn+bAgQPcd999vPbaay22UalUzZ4rFIp2zyHOEIROw1xXS31hFlrfMAD2HM/Ey1VHZJ+bbxttzYzY3thoFGzYnwo0jjOvCx6M/tJZzA31FnlNQfil0tJSZsyYQWhoKI888ggjR44kOTkZhUJBQ0PHXgYtCoLQadTlpgISGr9QCktrOZtWzMSh/habEMROp2ZyTAA/nM6hrMoANDYbScY6DJnnLfKagvBLLi4u3HnnncyfP5+5c+dSX1/PvHnz6NOnD1VVVTz22GMdlkU0GQmdhiEnBQCNTwhbD2YhSTA+updFX3N6bG+2HExn19FM7pgUijYgAplSTW3qSXRBrU/MIwjtZdiwYQwbNgyA3/zmN83W6XQ6vvnmm2bb/uTnk449/PDD7ZZHnCEInYYhOxmVmx9yjY49J7KICna77tzIN8vPw56BIe58d/gSJpMZuUqDNqA/tanxFn1dQeiMREEQOgVJkqjLTUHrG0ZyZhl5xTUWPzv4yYyRgRSX6zn2441quuBoGsryMZbmdsjrC0JnIQqC0Ck03ZDmF8oPp3NRKuSMiPTukNeO6eeJm5MN38ZlAD+7/FTctSz0MKIgCJ1CXW7j/QBq7xAOncllcJgHtjaq6+zVPhQKOdNGBHA6pYicompUTp6Nk+aI4bC7pZ+PYdWdmc3mG95HFAShU6jLS0emVHOp1pbicj0jB/h06OtPjglALpex+1gm0HiWoL+ciLle36E5BMvSarWUlJR066IgSRL19fXk5ORga2t7Q/uKq4yETqE+Px21ZyB7z+ajVMgZdgMzorUHFwctQ8I92Xsik3umhaMLjqbi6Fb0l85hGzq0Q7MIluPn50d2djZFRUXWjmJRSqUSR0fHGx7fSBQEweokyUxdQQZ2keM4dCKXQWHuHdZc9HOTh/lz7Hw+8RcKGRoejkytpTb1pCgI3YhKpSIwsP3vfO8uRJORYHXG0jykegNlai+KyvSM6uDmop8M6euJk72GnUcvI1OosAkcQG1qfLduXhCEnxNnCILV1eenA3C2VIdcXkNMv5tvLsqtzOdCcTp5VQUYzQ1olRq87Tzo5xGCu61rq/soFXImDunFxu/TKKs0oOszmNrkoxiLMlF7BNx0JkHo7ERBEKyuLj8dmULFgQyJ/oGu2OnUv+o4DWYTBy8dZVvKXjIrcgBQyBWoFSoMDXVNv/SDnP2ZFjKOUf5DUSqafwQmDvXn632p7IvPYvaQK5efioIg9AQWLQhbt27l3//+Nw0NDSxZsoSFCxc2W5+YmMgzzzyD0WjE29ubV199FQcHB0tGEjqhuvwMZK69SL9Qw72ze/+qY6QUp/Pv45+SU5lPoFMvfjvoDgZ698fT1g25XE6DqYHcqgIS8pPYnxHHu8c+YVPSDpYMms8g74im4/TytKdvbxd2Hs1kzrhg1J6B1KadxCl2Tju9W0HovCzWh1BQUMDKlStZt24dmzZtYv369aSmpjbb5sUXX2TZsmVs2bKFwMBAPvjgA0vFETopSZKoz0+nROEBwNAbbC6SJInNSTt5es9rGBrqeGzUA7wy5W9MDx2Pt70Hcnnjn7hSocTfyZdZ4ZN4bdrTPD76D0iSxMsH3uE/xz7F0FDXdMwpw/zJKaom6VIpuj6DMGRdwKSvbr83LQidlMUKQlxcHMOHD8fJyQmdTsfUqVPZvn17s23MZjM1NTUA6PV6tFqtpeIInVRDeQFmQw1J1fb4utvi627X9n3NJv597FPWntnIiF6DeX3a0wz1HXDd0VFlMhnRPpG8Pu1pbus7lX0Zh3ly1z8prC4GYOQAX2w0CnYfy0QXEg2SGX1Gwk29T0HoCixWEAoLC3F3vzINoYeHBwUFBc22eeKJJ1ixYgWjRo0iLi6Ou+66q8VxKisryc7ObvbIz8+3VGyhg9XlNw4XcTRPc0NnByaziVVHPmT/pcPM738Lj4y4F53K5oZeW6lQsiDqNp4a+zBlhgqe2v0vUksuYaNRMjLKlx8ScpBcg5Db2Im7loUewWIFwWw2N/ulJklSs+cGg4GnnnqKjz76iB9++IEFCxbw+OOPtzjOxx9/zMSJE5s9ftkXIXRd9flpSDI5WfWObb66SJIk/n38U+Ky4rlnwFzuiJh5U3MmRHn15YWJj6FRqnl+/xskF6cxKcYffZ2Jw4n52AQNRJ92Ckm68aEABKErsVhB8PLyanY3YFFRER4eHk3PU1JS0Gg0REVFAXDnnXdy7NixFsdZsmQJe/bsafZYu3atpWILHawuP51KlTtarYa+gS5t2ufr899x4NJR7oiYyezwye2Sw9fBi+cn/gVnrSMvfb8KlUMF3m627D6Wha7PYEw1FdTlpbfLawlCZ2WxghAbG8vhw4cpLS1Fr9ezc+dOxowZ07Q+ICCA/Px80tMbP2R79uwhMjKyxXEcHBzw8/Nr9vDy6thhDQTLkCSJurwM0gyODA73RKm4/p/jsezTfHFuK2MChjGv34x2zeNi48Sz45fjoLXnpQOrGDpIx9m0YqqdwwAZejH6qdDNWawgeHp6snz5chYvXsxtt93GzJkziYqKYunSpZw9exZHR0defvll/vSnPzFr1iy+/vprXnrpJUvFETohU2UxZn0laXonhvT1uO72hdXFvHPsY4JdenP/0IUWmVrTRefEM+MeQa1QEV/3DXJVHXvOlaHxDRGT5gjdnkXvQ5g1axazZs1qtmzNmjVN/z127FjGjh1ryQhCJ1b34x3KWQ2uLA29dkEwmU28feRDAP4Uex9qheXGOnK3deWJ0X/g2b3/h2PkGfbE2zFj9CDKD66nobocpZ2TxV5bEKxJjGUkWE1dfjpmZKjc/XFxuPYlxxuTtpNcks7S6LvxuMrQE+0pyCWAZSN+h0FZQoXTCbLUjQOi6dNPW/y1BcFaREEQrEafm06ByZGIsGsPZpdZnsPXid8yKiCGUQExHZQOhvoOYH6/mSjd8vgsLQmFrZO4/FTo1kRBEKxGn5tKVoMLg8Ku3lxklsysPrEOnVrHbwfd3oHpGs2PmI6LzJ9L8iPk+oehTz+NZDZ1eA5B6AiiIAhW0VBVhtxQSa7kRv+gqzcB7Uk7REpJOosHzMNe0/a7mNuLXCbnwSGLkYxa3jfmUVtfiyE7ucNzCEJHEAVBsIqfhrxWewSiUSla3aZcX8HaMxvp7xHKmN7DOjJeM1GBPriWjaBKqmOzuwN60WwkdFOiIAhWUXYpGbMEfn37XXWb/53bSp2pnqXRd1vkEtO2kslkTB84CGNuEKfsNRzKOGK1LIJgSaIgCFZRmpFMkdmBqL69Wl2fWZ7Dvow4pgWPw8fB+jcijhvcCyk/GHeTjq+0RgpyL1g7kiC0O1EQBKuQlWZSgDu9vVuf/+LThA3oVDbM6ze9g5O1zslew5Bwb8wZUTTIZPzn+FoxtabQ7YiCIHQ4Y3U5OlMlMrfeyOUtm4JO550nIf888/rNwE5ja4WErZsc4092qRNTapQkGoo5eLnl2FuC0JWJgiB0uKzziQC49wlrsc4smVmbsAFPWzemBo9psd6aovt64mSnwaE2EH+DkY9OfkGlocrasQSh3YiCIHS4vJTzAIQMHNBi3bHs01yuyOGOiFmoLDg8xa+hVMgZF+3HrhxX5hVWojfq+fDUF9aOJQjtRhQEocPV56dThgPunm7NlpslM1+e+wZfey9G+g+xUrprmxTjT1aDE84yB6Zgz6HME8TnnrV2LEFoF6IgCB3KZDJjr8/FYOfbYt2RrJNkVeZxe8QtTXMhdzYBXg6E9HLmXIM/ozIu42fvxX/j/0ddQ721ownCTeucnzqh20pLz8VFXo2NT59my81mM1+e20YvB2+G9xpspXRtMynGnx/KvFCYjCx0H0BRbSmbknZYO5Yg3DRREIQOlXGusXnFt2//Zsvjsk6QU5XP7REzkcs695/lmIG+ZONJvUJHr9zLjPIfypYLO8mvLrr+zoLQiXXuT57Q7VRcTgHALSi8aZlZMrPx/HZ6OfoQ4zfQWtHazE6nZliEL2frfKm5eIKFkbNRyBV8dFJ0MAtdmygIQodpMJlRlmehVzqi0F25Ie1UXiJZlXncGj6l058d/GRijD/x+l5IdbXY5F9mfv9bOJl3jhM5Z6wdTRB+ta7x6RO6hdSscrzlJeAa0Gz5pqQduOtciO2kVxa1ZkCIO0W6QOplGqqT4pgROgFfBy8+OvUF9aKDWeiiREEQOsy5C9l4KipxCQxtWnahKJXk4jRmhk1CKW991NPOSCGXMSEmiFMGP6ovHENhNnPv4DsprCnhm5Q91o4nCL+KKAhCh8lPbRwQzjHgSkHYdGEn9ho7JgSNtFasX23aiAASjIFQX0tt+mkiPMMZ4juATUk7KNdXWDueINwwURCEDmFsMNNQmAGA2isIaBzR9GTuWaaHjEejVFsz3q/i6miDS/ggaiUNled+AOCeAXMwmox8ce4bK6cThBsnCoLQIVIyy/CiGJPGAaWdMwBbLuxCo9QwLXisldP9ereMDuF0XS9qUo5jNtbhY+/JlOCx7Mk4RGZ5jrXjCcINaVNBePjhh4mLi7N0FqEbO5tWTC9lCVqfYADK9BUcyjzOxMDYTjWi6Y3qF+hCvn0/5KY6atNOAzC//wx0Si2fJmywcjpBuDFtKgiTJ0/m3XffZerUqXzwwQeUl5dbOpfQzSRdzMNTUYmtb+MdyrvTDmKSzEwLGWfdYDdJJpMROWo01WYNucf3AWCvsWNe/xkk5J/ndF6ilRMKQtu1qSDMnj2bzz77jHfffZeSkhLmz5/PY489xpkz4ppr4frqjSYqs9OQI6HxCqLB1MCutIMM8u6Pl72HtePdtLFD/DlvDkTKOo253gDA1OCxeNq58+nprzGZTVZOKAht0+Y+BLPZzOXLl7l06RImkwlXV1eee+453nrrLUvmE7qB5MtleFMMgMa7D0dzTlFuqOzyZwc/0aqVaMJiUUlGcuMPAqBSqFgYdRtZlXnsyxDNrULX0KaCsHLlSsaOHcv777/PjBkz2LlzJ0888QSfffYZa9eutXRGoYs7k9rYfyCzcUBh78L2i9/jaefOAK9+1o7WbkZPG0eJ2Y7cozublg3zG0SYWx++OPcNhoY6K6YThLZpU0EoLS1lzZo1rFu3jpkzZ6JSNU5cotPpeP3116+639atW5kxYwZTpkxptXCkp6ezaNEiZs+ezb333ktFhbh2uzs6m1ZMkLYCrXcQl8qzSS5OY2rw2C4zTEVbuDnZUuI6EJfqDErz8oDG/oWFUXMoN1TybcpeKycUhOtr0yfSZDIRHh7ebNmyZcsAGDVqVKv7FBQUsHLlStatW8emTZtYv349qampTeslSeLBBx9k6dKlbNmyhb59+7J69epf+z6ETqrOaCLtchGuUikaryC2X9yPRqFmfOAIa0drd/0nz0Qukzi9Y2vTsnD3PkT7RLL5wk6q6qqtmE4Qrk95rZXPPvssBQUFxMfHU1pa2rS8oaGBrKysax44Li6O4cOH4+TkBMDUqVPZvn07Dz30EACJiYnodDrGjGmcN/eBBx6gsrLypt6M0PlcyCjFg1LkmKl39+GHCxsZ23s4tmqdtaO1u14hIRxT+2CTdZRaw2/RaRvPpO+OvJXHdrzIxqQdLB44z8opBeHqrlkQ5s+fz8WLF0lOTmbq1KlNyxUKBQMHXnuY4sLCQtzd3Zuee3h4NLsqKTMzEzc3N5588kmSkpIICgri6aefbnGcysrKFoUiPz//2u9K6DTOpBXTS9X4Y+KwsQyjydilb0S7HueB41EcW8uB3XFMm9n4Pv2dfBnTexg7Lu5nRsh43GxdrJxSEFp3zYIQGRlJZGQkI0eOxNPT84YObDabkclkTc8lSWr2vKGhgWPHjvHZZ58RGRnJG2+8wSuvvMIrr7zS7Dgff/wxq1atuqHXFjqPs6nFjHaoBqUdu3NO0s89BH+nltNndhdBoyaTfux/lJzci37ySGw0jR+xOyJmcijzBF8mbuPBmEVWTikIrbtmQXjkkUd48803ue+++1pdv3Xr1laXA3h5eXHixImm50VFRXh4XLnm3N3dnYCAACIjIwGYOXNmU7/Ezy1ZsoQ5c+Y0W5afn8/ChQuvFV3oBPR1DaRklrHQu5RUL2+KakpYNGCutWNZlMLGHlmvAURcPs83By5y++S+ALjbujIleAzfXdzHrLBJ+Dl6WzmpILR0zYKwdOlSgFabcq4nNjaWt99+m9LSUmxsbNi5cyf/+Mc/mtYPGjSI0tJSLly4QHh4OHv37qV///4tjuPg4ICDg0OL5ULnl5RRCuYG7OoK2aANxFXpzFDfAdaOZXE+I6aRn3WS5IN7qRkVjK1NY1/C3H7T2Zcex+dnN/PYqAesnFIQWrrmVUYREREAxMTE4O3tTUxMDLW1tRw/fpy+ffte88Cenp4sX76cxYsXc7pPPo0AACAASURBVNtttzFz5kyioqJYunQpZ8+eRavV8s4777BixQpuueUWjh49yhNPPNF+70ywurNpxfiqKihSSCQZK5gcPBpFF5rz4Ney6TMQSefMIHkSWw6kNS130NgxK3wyx3MSSClOt2JCQWidTJIk6XobPfPMM0Bj882SJUsYPXo01dXVvP322xYP2Jrs7GwmTpzInj178PPzs0oG4fr+8uYB+hnPUWl/lGPO9vx79ks4anvG2V7ZgS8oO7ief9bezj+fmIOTvQYAg9HAw98+i4+9J8+NX96sX00QLO16351tug/h3LlzPPfcc+zatYs5c+bw8ssvk5MjhvYVrq7WYORidjmBduXEO9gwwj+6xxQDAPuBE0EmI1p+gbU7LjQt16q0zOs3naSii5wSA98JnUybCoIkScjlcg4dOsTw4cMBMBgMFg0mdG3nM0oxmyWyVfnUyWVMDxlv7UgdSungii44mtF26ew+kk5G7pW78CcFjcLT1o3Pz2zCLJmtmFIQmmtTQfD392fp0qVkZ2cTExPDo48+SlhYmKWzCV3YubRiNAoTR9QGApS2BLv2tnakDucwaAqahhqibXN5f/M5fmqdVSqU3Bk5m8sVOfxw+biVUwrCFW0qCC+//DIzZ87k008/RaVSMWTIEF566SVLZxO6sDOpxYQHllCkVjDJM8racazCps9AFA5uzPbM4kxqMYfP5jWti/WPJtC5F+vPbsFoMloxpSBc0aaCoNPpGDJkCJWVlSQmJhIVFUV6urhKQmhdjd5IWnY5dU6XsG0wMypsnLUjWYVMrsBh0GTsyi4y2LOB9zaepUbf+OUvl8lZGDWHotpSdqYesHJSQWh0zfsQfvLmm2/y3//+F1dX16ZlMpmMPXv2WCyY0HWdzyhBUunJlJUyrrYBnVvPvRLMYdBkyn/4ikWBOTx6NICPt53nD/Mb78WI8upLlGdfNpz/jvGBsejUNlZOK/R0bSoImzdvZufOnTc8fIXQM51JLUbl1Tj44RgbL2TdaJjrG6WwdcQuYjTV5w8xJ3YEXx+6xNjBfvQPavxxtSDqNp7Y9TJbkndyV+StVk4r9HRt+qR6e3uLYiC02dm0fFQe2fSvrcfLK9TacazOYegtSMY6Zrpn4eGi4631pzDUNQAQ5OLPSP8hfJO8hzK9mA9EsK42FYQRI0bwr3/9i/j4eBITE5segvBL1Xojlw3JmOX1jCivQeMdZO1IVqfx7I02IIKa0zv40+1R5JXUsGbzuab1d0XOxiSZ+TJxmxVTCkIbm4w2bNgAwPbt25uWiT4EoTWJacUoPDPxkNsRpC9E7dXH2pE6Bceht1Dw1T8JNKUzb3wIX+29yOBwD0ZG+eBp587kPqPZmXqAmaET8HHwsnZcoYdqU0HYu1dM/ye0zcGLichtKxmv8kOutkHlIr7cAHQh0SidPKg4tpWF97xAwsUiVn1xmhA/JzxcdMzrN539GYf5/OwWHh15v7XjCj1Um5qMampqeP7551myZAnl5eU888wz1NTUWDqb0AWdKT+O3KxiUFklGq+gHt2h/HMyuQLHmFnUZSdjzLnAX+6JxixJvPjRMQz1DThqHZgdPpmj2afEwHeC1bTp0/rCCy9gb29PSUkJGo2G6urqpgHvBOEn2aXF6LVZBNn0R1ZwWfQf/IL9wInIdQ6Ux23Ax82OvyyMJiO3gne+TECSJGaGTsRRY8/aM5tow5iTgtDu2lQQkpKSWL58OUqlEhsbG1577TWSkpIsnU3oYr48vRuZXGK6ZziSyYhG9B80I1dpcIyZhT7tFHX56Qzt58XCaeHsP5nNxv1paFVa5ve/RQx8J1hNmwqCXN58M5PJ1GKZ0LM1mE3EFx1HqnQnQlMPgFqcIbTgGD0VmUZHeVzjhRp3TAxl5AAfPvwmkQOnspnYZxRedu58lrABk9lk5bRCT9Omb/WhQ4fy6quvYjAYOHjwIA899BDDhg2zdDahCzmWfZp6avGTRdCQn45co0PlIqaJ/CW51hbH6GnUJB2hviQXmUzGn+8eTP8gV1Z+fpLEtFIWDZxHdmUeu9IOWjuu0MO0qSD85S9/QafTYW9vzxtvvEF4eDh//etfLZ1N6EK2Je/DbLBhWEAUdbkX0fgEiw7lq3CMmYlMqaI8biMAapWCFb+Nwcfdjhc/PIaT2Z9IzzDWn9tKVV21ldMKPcl1P7G7du1i0aJFvP/++2RnZ2Nvb8/gwYPRaDQdkU/oAi6VZXOxNA1ToT8DejtSX3gZjU+ItWN1WgpbR+wHTqL63PcYy/IBsNOp+fvSEdjpVDzz3mEmeE+j1qjny3PiZjWh41yzIHz33Xe8+uqr3HPPPXz55Zd89tln3Hbbbbz44ovs3LmzozIKndz21P3IUaKoCMBfXQaSWRSE63CKnYtMrqDs4BdNy9ycbHjpwZHYaJW881k6w7yGsTPtAFkVuVZMKvQk1ywIn3zyCR999BEzZ84kODiYPn36cNttt/Hee+/x4YcfdlRGoROrrqvhh8vHUFX1IiLAC2N+KgAan2ArJ+vclPbOOAyZTvXZA9QXZTUt93K15aUHR6JRKzm+1wG1XM3Hp74Sl6EKHeKaBaGmpgYfH58WywMDA6mrq7NYKKHr2JdxmHqTkYpLPgwMdacuNxWlgxtKO2drR+v0nEbMQabWUnbgf82We7na8vIfRqJV2FKX2YczBUkcz0mwUkqhJ7lmQVAoFFddJ36xCGazmR2p+/HW9kLS2zMgxJ26nIuiuaiNFDp7HIfNoubCEerymt+d7OVqyz8fGoVTfRiS3p73jn2OwSjmMRcsS1wGIvxqp/ITKawpwbYmBCc7DX72Eg0VhWh8RUFoK6dhs5Db2FG677MWP7I8nHX866ExeNQMo8pYyau7P7dSSqGnuObgdsnJyQwePLjFckmSqK+vt1gooWvYfnEfLjZOZJ20JSrEjXrRf3DD5BodzqNup2TXh+jTTqILjm623tFOw2v33cqy9Zc5U3GcNTtCuW9KLDKZzEqJhe7smgVh165dHZVD6GJyKvNJyE9iWu+pbKwyMjDEnbrcOJDJxZAVN8ghehqV8Tso2f0RNoEDkCmafyxtNEpenX8/f9jyNDuyt1L/lY4H5g5AoRAn+EL7uuZflK+v7zUfQs+1/eJ+VHIldrWNX/4DfuxQVrv3Qq7WWjld1yJTKHGZuBhjSS6VJ1u/nNtJZ8cDw+9GblfBrowDvPDhMfQ/zromCO1F/MQQblhtvZ79l44Q6z+EC+k1+LjZ4u5kQ11uquhQ/pV0IUOwCYyi7MB6TPqqVrcZHTCUwd4R2PRO5dSldJ545wdKKvQdnFTozixaELZu3cqMGTOYMmUKa9euvep2+/fvZ8KECZaMIrSj/ZcOU9dQx+Q+YzmXVsyAUHeMpXmYDdWiIPxKMpkM10m/wVxXS+n+dVfd5vdD78FGpSEgJp284kr+8tZBLuVVdnBaobuyWEEoKChg5cqVrFu3jk2bNrF+/XpSU1NbbFdcXMw///lPS8UQ2plZMrP94n7CXINoqLJHX2dq7D/IvgCA1i/Uygm7LrVHAI5DZ1B1cieGH/89f8nZxpH7ou8iT5/DlNkmzGaJx1cd5HRKYQenFbojixWEuLg4hg8fjpOTEzqdjqlTpzabk/knK1as4KGHHrJUDKGdnc47T351EdNCx5GQUoRMBlHBbhiyk5FrbVG5+Vk7YpfmPPYuFA5uFH37HySTsdVtYv2HMKJXNHuy9vDwkt54OOt4bs0Rdh+73MFphe7GYgWhsLAQd3f3puceHh4UFBQ02+aTTz6hX79+DBgw4KrHqaysJDs7u9kjPz/fUrGF6/ju4j6cbRwZ5jeY0xeLCPZzwk6nxpB9AY1vmBjh9CbJ1Ta4Tb0PY1EW5Ue2XHW7pdF346x15MOzn7Bi6SAig914c/1pPtueJG4aFX61a152ejPMZnOza6UlSWr2PCUlhZ07d/LRRx9d8wv+448/ZtWqVZaKKdyA3Mp8EvLPc0fELOrrzSRfLmPu+GBM+iqMxdnYRYyxdsRuwTZ0KLbhwyk/+CW2IUNRe/i32MZOY8ufY5fy9N7X+PDMOp6+937+8/VZ1u9KoaCklmV3DkSlvPpIA4LQGov9nPPy8qKoqKjpeVFRER4eHk3Pt2/fTlFREfPmzeP++++nsLCQBQsWtDjOkiVL2LNnT7PHtTqoBcvZfvF7lHIlk/qMIuFiMSazxKBQDwzZyQBo/cKtnLD7cJt2P3KtjsLNbyA1tN50FOzam8UD5nEy9yzfXtzDw3cM5J7pjVNyPv3eYapqxc2jwo2xWEGIjY3l8OHDlJaWotfr2blzJ2PGXPkFuWzZMnbs2MHmzZtZvXo1Hh4erFvX8uoKBwcH/Pz8mj28vLwsFVu4ilqjnv2XDhPbKxonrQPxFwqw0SjpG+hCXXYyyBXiDuV2pLB1xO2WP1BfeJnS71u/6ghgWsg4RvSK5vMzm4nPPcOdk8L4y8Joki+X8dhbB8gtFhPsCG1nsYLg6enJ8uXLWbx4MbfddhszZ84kKiqKpUuXcvbsWUu9rGAh+zMOY2ioY1rIOCRJ4mRyIQND3VEq5I39B15ByFVi0qT2ZBsyBPvBU6g4shV9xplWt5HJZDwYs4ggZ3/ePPxf0kszGTvYjxceiKWyxshf3jzI+YySDk4udFUW7QGcNWsW33zzDTt27GDp0qUArFmzhsjIyGbb+fn5sXfvXktGEW6CWTKz4+L3hLgGEuzam6yCKorK9AwO80AyGanLTUXrF2btmN2S68QlqNx8Kdi0EmNF65eWapUaHh/9IPYaO/558F2KakroH+TKa8tGY69TseI/cRw4ld3ByYWuSFwSIlzXydyz5FUXMiN0PADxFxq/mKLDPanLz0BqqEcj+g8sQq7W4jn/cTA1UPDlvzAbW5+HxMnGkSdG/4E6Uz3P73uDktoyfNzteHXZGEJ6OfHqZ/F8sTtFXIEkXJMoCMJ1bU3eg5vOheF+jSPfxl8owN/LHndnm6YbqESHsuWoXX3wuPVP1BdcomjrKiSzqdXt/J18WTF2GZX11fx930pK9eU42Kp54YFYxg3249Pvknhr/WmMDeYOfgdCVyEKgnBNaaWXSSq6yIzQ8SjkCvR1DSSmlxId7gmAIesCSicPlPZihjRL0oVE4zJxETVJcZTs/O9Vf+kHu/bmyTEPUW6o5Nm9/0deVSEqpYI/LxjMXZPD2H08k+fWHKZa3/qVS0LPJgqCcE3fJO/GRqllQtBIAM5cLKLBZCY63ANJMmPIPI/Wv7+VU/YMTsNvxXH4rVTGb6ds/9qrFoUwtz6sGLuM2vpaVux5lZTidGQyGQunhbP87kGczyjhr28fIL+kpoPfgdDZiYIgXFVxTSmHs04yMWgkOpUN0Nh/YKNR0C/QFWNRFmZ9FTYBoiB0FJcJi7AfNJnyuI0Ub1991eajULcgXpj0V3QqG/6+/w32ph9CkiQmDPHn+ftjKa2s47G3DpJ8ubSD34HQmYmCIFzVdxf3ATAjtHEkWkmSiL9QQFSwOyqlHP3lRABxhtCBZDIZbtN/j+OI26g6uZOCr1/DXFfb6rbe9h68OPExwlyD+M/xz3j76EfojQYig9149eHRaNQKnnz3EIfO5HbwuxA6K1EQhFbVGvXsTv+B4b0G42brAkBmQRWFZXqi+zb2H+gvn0Pp6IHKyeNahxLamUwmw3XCIlwn/5baiyfI/uAx6nJbjiQM4KC1Z8XYZdwZMYtDmcd5dPs/OJGTQC9Pe15/ZAyBvo7885PjfL7jAmazuAKpp7PYWEZC17Y3PQ690cDM0IlNy46eaxxzKqafZ1P/gS5kqLUi9niOMTNRewVRuHElOR/9DYch03AedTsKnUOz7eRyOfP6zyDCM4zVJ9bxrx/+wxDfASyIupUXHxzJu18lsG5nMsmZZTy6MBp7nRpJkjBVlWAszcNUU46ppqKxeUomQ67WoXRwRe3mh9LR/SrphK5IFAShBZPZxHcpe+nrHkywa++m5ccS8wnp5YSrow11BZdE/0EnYOPfj16/f4PS/euoPP4dVaf3YD9wIvaR41F7BTYbUDLMrQ//nPIk3yTvZsP573h0+xlG+8cwb/J4BjpXceRgHNtWbmeopxFFRQ5mw/WHvVDYu2ITOAD7iNFoA/ojk4sB9boyURCEFo5mn6aotpTfDL6jaVlppYHkzDLumd54v4Eh88f+g4B+VskoXCHX2uI2bSkOQ6ZTHreByvgdVB7/FoWDG1rfUFSuvijtnJGp1EhmM+MMdQyy68f2inQOXDrKgctHCa6tZ7h7LX2qzVzOccaxV3+CoyJRu/mhsHNGoXNEplCCZMZkqKGhspj6gksYspKouXCY6jN7UTp54jRyLvaRY5EpVNb+ZxF+BVEQhGYkSWJz0g687T2I9rkyxMixxMbmouH9vQHQX05E6eSBylH0H3QWajc/PGYvw3Xyb6m5cBT9pTPU5aRQc+EISM1vRpMp1cxwcGOCmz/HbZV8r8rnM50aW5UOVY0vBWlODDI78sidITjba5vtK9faonLywMa/H45DZ2A21lF78QQVRzZTvO3fVBzejNv0+7Hp3XyIGqHzEwVBaCYh/zwZ5Vk8OHQR8p9NdnM0MR8vVx3+XvZIZhOGzER0ITFWTCpcjcLGHodBk3AYNAkAyWzCVF2OZG4AZCi0tsg0uqbmpBDgTrOJ0/nnOXT5OMdzz6Dpe5FEYwIPrtvH9IgY5g2NxU5t2+rryVUa7PqNxLZvLLWp8ZTs+pC8tc9hP3ASrlN+JwY97EJEQRCa2Zi0A1cbZ0YHXPmy19c1kHCxiOmxvZHJZBjyMjDrq9EFRlkxqdBWMrkCpYPrNbdRyBVE+0QS7RNJXUM9J/POciDtJCdzz7EtewPbsjYS4hpEjF8UQ3yj8HVoOQS9TCbDNmQINr0jKTv4BRWHN1GXm4LH3MdQu/pY6u0J7UgUBKHJhaI0koou8ptBt6NUXPnTOJVciLHBfKW5KP00IMNGFIRuSaNUM6JXNCN6RVNnNLJm1w/sS4nnoqGIi6UbWXtmI772XsT4DSTGbyBBzv7NOq/lKg2uExZh49+fwi1vkvvR3/C6429oe4nxrjo7cR+C0GRT0nbsNXZNw1T85GhiPvY6Ff0CG+9H0KefRu0VhMLW0RoxhQ6kUal4aMZ43rj79wRWz8Rweiyu1UPQym3ZfGEnf9v1Cg9te5qvErdRXNv8rmdd8GB8f/cvFDp78tb9nZrkY1Z6F0JbiYIgAHC5PJuTeeeYETIerfJKm6+xwczRxHyG9vNCoZBjrqvFkJOCLmiAFdMKHa2Xpz0v/2EUf7x1OBWXfEjcHUKMtIglUXfjbefBF+e+4Y9bV/DS92+TkH++aZwllZMnPkteQu0RQMGG10RR6OREQRAA2JS0A61Sw9SQsc2WJ1wsokZvZNSAxjZg/aWzYDZhEzTQGjEFK5LLZUwd3pvVT05i9ug+HDhRyMefVtFPms7Kac8xt990Llfk8OL3b/O3Xa9wJOskZsmMQueA94Jn0HgFUbDhdWpTT1r7rQhXIQqCQH51EXFZ8UwJHtviSpKDp3Ow1SoZGNp4eWlt+mlkai1av1BrRBU6ATsbFffdGsE7f51ARB9XPvzmPM+9cwZ/onl7xvP8fshCao16/i9uDX/b9QrnC1OQa3R43bUCtXsvCr76V9M8GkLnIgqCwMbz21HKFMz8cRC7nxgbzBw9l8fwSG9UysY/FX16AjYBkeLGIwFfdzueuXc4f79/BGqVgpc+Os5zq48TqI3kjenP8dCw31BpqOa5fSt57dB7lEtGvBc8g8LBlfwv/4mxLN/ab0H4BVEQerj8qkK+v3SEyX1G42TTvJP4dEohNYYGRg3wBcBYmkdDeYFoLhKaGRzmwVt/HseD86K4lFfJn1bu552vzhDpMoA3ZjzHXZGzOZ2XyJ+3P8++vDN43fkkSGby17+ISV9l7fjCz4iC0MN9ff47lHIFt/Wd2mLdDwm52NqoGBDSOIBZzcUTAOiCB3VoRqHzUyjkzIgNZPWTk7h1TB/2HM/k96/sZsv3l5gZMoXXpj1NkLM/q0+s5V9nv8bm1mUYywsp3Ph/V53TQeh4oiD0YLlVBRy4fJQpfca0ODswNpg4ci6PERFXmotqLx5H7eGPysnTGnGFLsDORsW9sxv7F6KC3fnk2yQe/NdecrLNPD3uEe6Lvouk4lSePvcFxWNuQ59xhrLv/2ft2MKPREHowb5O/Ba1XMXsvlNarDuRVECtoYFRAxuvLjLpqzBkJonhroU28XW3Y8XvhvHC72NRK+U8t+YIb/7vNCN8RvDypMexVel4Pfsgh/tFURa3gZqU49aOLCAKQo+VXZnHD5nHmRoyFietQ4v1e09k4WyvYeCPzUW1qSdBMouCINyQAaHuvPnncdw+MYT9J7P547/2kp+j4OXJjxPbK5rN9fl8HeBD3pa3MJbmWTtujycKQg/1VeK3qBVqZodNbrGuorqOE0kFjB3sh0LxY3NRynEUds5ofPp0dFShi1OrFCye0Y//e2QMTvYa/vHfo3yy7SIPDlnCHRGzOKFqYLWHDWkbX0cyGa0dt0cTBaEHulyezeHMeKaHjMNBa99i/cHTOTSYJCYM6QWA1GCkNv0UupAhyGTiT0b4dfr4OfH6I2OYPSaIrQfT+euqH4j1HMOfRtxLjlbNW+oKUvd8ZO2YPZr4dPdAaxM2olPbMDu85dkBNDYXBfk4EujT2NGsv3QGqd6ArWguEm6SSqlg6a2RPPXbGApKavnzGwewNQTw9PjlVKvVvFp4nPSk760ds8eyaEHYunUrM2bMYMqUKaxdu7bF+t27d3Prrbcye/Zs/vCHP1BRUWHJOAJwtuACp/PPM6fvtFbHt88qqOJiVjnjfzw7AKg+H4dcoxOjmwrtZniENyuXj8XVUcszaw6TliLn2Ql/xqhQ8MKpz0nPE3cyW4PFCkJBQQErV65k3bp1bNq0ifXr15Oamtq0vrq6mueee47Vq1ezZcsWwsLCePvtty0VRwDMkpnPEjbgpnNhWsi4VrfZeyILuVzG2MGNN6OZG+qpSTmGLmw4MqW4O1loP16utrz68GgGh3nwnw1n2Huolqdj7kUuSTx/4G0ySjOtHbHHsVhBiIuLY/jw4Tg5OaHT6Zg6dSrbt29vWm80Gnn22Wfx9Gy8pj0sLIy8PHGVgSXFZcaTUZbFXZGzUbcy9ISxwczuY5kM7evZNG2iPu0UUl0tdv1iOzqu0APotCpW/G4Ys8cEseVgOhuPyvir/wRURiP/2PM6meU51o7Yo1isIBQWFuLu7t703MPDg4KCgqbnzs7OTJ7c2IZtMBhYvXo1kyZNanGcyspKsrOzmz3y88UYKDfKaDLy+dnN9HbyY1RA630BR87lUV5dx7QRvZuWVZ8/hNzGXsyPK1iMQi7jvtkRLJrel/3x2fwv2Z8/yr2Q1xt4fu9KcirF572jWGzGNLPZ3GwWJUmSmj3/SVVVFX/84x8JDw9nzpw5LdZ//PHHrFq1ylIxe4ztF7+nqKaE349d1myu5GbbHL6Eh4uOwWGNI5s2Tp4ej13EGGQKMbmeYDkymYw7JoVip1Pxnw1noHcs95vX8557Lc/vW8lzEx7F297D2jG7PYudIXh5eVFUVNT0vKioCA+P5v9DCwsLWbBgAWFhYbz44outHmfJkiXs2bOn2aO1Dmrh6sr0FXyVuI1B3hFEefVtdZusgirOpBYzbXgAcnlj4a5NOYZkNIjmIqHDzIgN5NEF0cRfquOkeRz3ZRdjrKvl+X1vUFxTev0DCDfFYgUhNjaWw4cPU1pail6vZ+fOnYwZM6Zpvclk4oEHHmD69Ok89dRTrZ49ADg4OODn59fs4eXVcoJv4erWntmI0dzAbwbdftVtth+5hFIhY1KMf9OyqoS9KB090Ab074iYggDA2MF+LF8QzXfZTpTL+vG7y4XU1tfw4vdvU1lXbe143ZrF2gE8PT1Zvnw5ixcvxmg0Mn/+fKKioli6dCnLli0jPz+f8+fPYzKZ2LFjBwARERFXPVMQfp3k4jQOXDrKbX2nXvWUu9ZgZPexTGIjfZo6k43lhegzzuI85g5xM5rQ4cYN9sNsNvPu/4w85ZbNb4r0vO9WzCsH3uGZcY+gVWmtHbFbsmjD8KxZs5g1a1azZWvWrAEgMjKSCxfEtcaWZDab+W/8elxsnJjbd9pVt9t5NJNaQwO3jr0yLEX1mf0A2EWNs3BKQWjdhCH+mEwS739dwXLTdn7jHskHZZm8dmg1j49+EJWYpKndiZ9+3dju9INklGexeOC8q/6iMpnMbDmYRv8gV0L9nQGQJDNVZ/ZiExiFylF05AnWM3lYANNmjefb2ij6JCewxGsIZwqSeOfox5jNZmvH63ZEQeimSmrLWJuwiUjPMEb0ir7qdofO5FJUpmfOz84O9BlnaKgown7AhKvuJwgdZdboIFxGzSXN6EGfQ7tYEDqZuKx4/ntqPZIkWTtetyIKQjckSRLvx3+OWTJz/5CFV+2wlySJDftT8XW3ZWi/Kx31Fce2obB1wjZsWEdFFoRrWjitH9lhd2FsMBN08BCzwyaxM/UAXyZus3a0bkUUhG7ocFY88blnuTNyFp527lfd7nhSAWnZFcwZF9J0qWl9SS76tJM4DJ4qhqoQOg2ZTMbiO8dwxm0qdlWXCLtQyrjAEXyVuI3vUvZZO163IQpCN1NZV81/T66nj0sAM0Ku3uQjSRLrdlzA00XHxKFXBrKrPL4NFErsB7ecRU0QrEkhlzHvvkWkq8NwSt3BUEMfhvoO4MNTX3Dg0lFrx+sWREHoRiRJ4r3jn1Fj1PPA0HuQy6/+v/doYj5p2RXcNTkU5Y+T4JgMNVSd2Y9dv1Eo7Zw6KrYgtJlapWTE/X+lVm6Lef8aJrtMI8IjjHePfcKJnDPWjtfliYLQjexNP8TxnATujryVACe/zWDuugAAFJRJREFUq25nNkt8viMZb1dbxkf/7OzgxHdIRgOOMbd0RFxB+FVsHZ3wnfsn3BRVXPjyPW4PWkCgcy9Wxq3hfGGKteN1aaIgdBN5VYV8dOpLIjzCmBk28Zrbfv//7d15eFN1vsfxd/ak6U6bhrZStg6FAhYsFlFAvCgoO4PKuMCotyyjojjTuaxecNDeURQvVLhcdEbGYZwrKDIwCggiICClrC2UAkUWu6TpGppmz7l/AMVSaDtok1R/r+fhKSf5nfRzvn2efHPOyfmdw99xtriGiQ90q79FptdhoyZ7I7oufdEYO/sisiDcsqikPmj6jORO5Sk++fPHPN3zKQzBUfxx9wrOimmzb5loCD8BLo+Lpd/8CaVCybNpk286eR2A3eFm9T9P0PW2cO7te20vwnLwc7y2WiIGPuKLyILwg8UNewLaJTBGuZvlf9rDs32nEKwO4tVdy8QMqbdINISfgPcPr6Ww8jzT+j1Bu6CIJseu23Gaiho76WN61n+zyOuwUb1/I7rOfdDGJfoisiD8YDKFiviHf4dOLWecZzNZq3N5KW06cmQs2rkUs7XC3xHbHNEQ2rivvt3HF4W7GZ30AGnxfZoca6qsY/2OMwxKiaNHp3b1j1fv/QRvnYWIwRNbO64g/KjU7WIxjnuReEUFd1m3sfLvZ/ndgN9gc9lZuGOJaAr/ItEQ2rCzlRdYlfM3ehq68ateo5scK0kS76w9glwuY/LIHvWPu2vM1GRvIjh5INrYrq0dWRB+dPrEVCIGPko/dSEG014+3FDC7IHPY3XWsWDHEspEU2gx0RDaqIq6Kl7/egWh2hBevOsZFHJFk+O/zLnI4VNmJo/ogSEiqP7xyh2X7y0ROeTxVs0rCK0pfOAEghJTGac/yKUzR/n4n2ZmD5pBncvGwi/foqy23N8R2wTRENqgOqeNzF3vYHPZmTXwN4RqQ5ocX1Fj490NeXTvGMlDAzpde52zR6g9vpuwtNEow25+RbMgBDqZTI5h9AzUUfFMi9jN+RMnWLepjNn3PE+d286CHUsovVTm75gBTzSENsblcfHm3pUUWUr47d1TmrzeAMDjlXhzzSFcHi8zHk25diLZaaf8s5Wo2sUSfs8vfRFdEFqVXKun/cS5qIOCeCl6J/nHz7B2k4k5A2fgcDuYv30xZyvP+ztmQBMNoQ253AxWkWsqYFq/J296O8zvW7v9FLmF5Uwb15t4w7U9icoda3DXlBH10HTkSnVrxhYEn1GGRtF+4jzUMjezY78m9/g5/r6hlPmDX0KtULFgxxKOleb7O2bAEg2hjXB73CzZ+y6HinP59zt+xeBO/ZtdJyffxIdbTnJv3/gG8xVZTx3AkvMZof1GoOvQo4lXEIS2R21IwDjh9+icVbwcv5vjJ8+x4m+FzLr7RWKCo8nclcXOb7/xd8yAJBpCG2B3O3hjz0pyio/xdN9HeaDroGbXOV9i4fUPcugYG8azE26vnwLbXWPGvCkLtbEz7e57srWjC4Jf6Dr2Iubh/0BnL2dBhz1cvGjiv97LZUbf6XSPTuSd7NX85fA6PF6Pv6MGFNEQApzFfomFO5ZwpPQ4U1IfY3jivc2uU1ZVx8L3vkGnUTD/6TS0mst3SvU6bJR+lInk9RIzbqaY3lr4SQvqnELMw/+BxlrGwvidOKvNzFuezS87Ps7wrvey6dR2Xtu1jEuOWn9HDRiiIQSwC9VFzN3+Bhdrism4eypDuwxsdp2KGhvzVuylzuZi/jP9iQrXASB53JjWv4nTfJGY8b9DFRnb2vEFwe+CuvTBOHEuSkc1s6O+IE5ZzbwV3xDn6s/0fk+Sby7k91tf47iYFA8QDSFg7Tq3nznb/ojT7eTle18kNe72ZtcprbAyZ/keqmvtLJhyF13jL09hLXnclH26BFvhYaKGpxPUufnXEoSfCl3HXsQ++QcUchnp6k2MiK8ka+0R8nKCmDdwJmqFild2vM1fj36Cy+Pyd1y/Uvo7gNCQ1VnH6sPr+OrcPnpEJ/LiXc8Qrgtrdr1TF6r4w3v78Xi9LEwfQFJCJABel4OyDf9NXcF+Iof+mlBx4xvhZ0gT05G4pzIxrXuD+0r+wS+S7uatbC+5hSE89+h09ld+yT9OfsHhkuOk3/ErkqJ/nlftiz2EAHKwOJffbv4Du87vZ3yP4cy/94Vmm4EkSWz55hyz3/katVrBH58bSPdOl5uBu7aKkr/+J3UF2bS7/ynC00b5YjMEISApQ6OInbSI0DuGE1+2hze6fE2Ep5z5y/cjL+7Ni2lTqHPZePnLN1m+/y/U2C3+juxzYg8hAFysKeYvRz7maOkJbguLJeOeaXSJTGh2vZpaByvX57L7SBEpv4jmpcf6EhGiBaDu26OYN2bhtdUS88sM9Elprb0ZghDwZEoVUcPT0SYkU755FVMVn3C6y9387y43uw7refyhZyhTHeOfp7ax77tDPJQ4hFFJQwlW6/0d3SdEQ/CjC9VFbDi5lT0XctApNUxKmcDwroNRKpr+s3i8Etuyz/P+phPYHG6efLA7E+5LRC6X4bFbqdr5IZacz1G1i8P4yGxxwxtBuE5w9wHoOiRT/sWfSDy+i9fjItjuvoOsv9voYAzj8cFTOe3OZn3+Zrae2cnQLgMZljiYqKBIf0dvVaIh+Jjb6+FISR7bCr/mUEkeGqWGhxKHMLbHcEI1wU2u6/F42Xm4iI+2FVBkttKzSzumj+9NB2MoXpeDmoPbqdr9EV5bLaGpDxF53xPIVRofbZkgtC0KfRgxY2di6/sAlds/YGjxNgbHR7HT3p0//188BkMnRvRLplR5lH8UfMHGgm3cGZ/CfZ3upldMt2YnlGyLREPwAbfXw0nzGXKKj7HnQg41dgthmhAeTh7B8MR7CWmmEZgq69iWfYFtBy5QXm2jY/tQZk3ux4Be7XHXlFG58zMsh7bgrbOgTUim3dCn0Bg7NfmagiBcpuuQTOyvM6kr2E/1vk8ZWryb+ww68unMl1vi+E4ex+3J3VDFXORY6RG+uXiIUE0wd912B6lxveke1RX1T2T6l1ZtCBs3bmTFihW43W4mT57M4483nGI5Pz+fuXPnYrVaSU1NZeHChSiVbb9HuT1uvq2+SEH5WU6VnyXXlI/VZUMpV5LSPpn7Ot1FSvueKG/yCcPh8lD4XTWHCso4cMLE2aIaZDLo083AtLHJ9G5nx3E+m+LV2TiKCgAISkwlrP9otLf1qL8qWRCElpHJZOiT+qNP6o/9uwIsh7bQs2A/yaHHcSj05J81knfciFyeSnwXOTJNCdsL97DlzE5UciXdoxPpYUika2RHOkd2aLPnHFrt3ddkMrFkyRI++eQT1Go1EydOJC0tja5dr32dKyMjg0WLFpGSksKcOXP46KOPeOyxx1or0o/K4XZSZaum0lZDpa2K4ksmvrOUUmQppeRSGW6vG4BofTvujO/DHbG96B2ThFalrX8Np8uDqbKOkgorxWYrxeZaTl+s4ttiC16vh3C5ndvjFIy500uivhal5SiOrYWU2C9fWak2JBA55An0yXejCjP4pQ6C8FOjje+GNr4bXpeDulMHsJ4+QN9vj5GiKASgrkTH+fMRXPCGc16vorYdnHEXccx0bdK8KF0k8WHtiQ0x0D4khpjgaCJ1YYRrQwnW6Ju877k/tVpD2Lt3L/379yc8/PLFUcOGDWPz5s0899xzABQVFWG320lJSQFg/PjxLF26tFFDsFgsWCwNv/5VWnrrN9CWJImjpflYHJfweD24vR48kge314376rL38rLD7aTObcPmslPnuvzT5rJjcVzC6rI1eF2ZTEaMPgod4STqUgiTxRAsGcCtw3HOw65THrY4jlJjdVBT66Cm1kkv6SRxyiq0Mic6mYuuCjd3qD2ERNvRuC8hk7xgBc6AW6lGFnUb+m5paDv2RJfQC2VI0/dPFgTh1slVGoKT7yE4+R4kyYvTdA77xZMEl35LSPEZkipOIvN4oQwoA6tMRqFGxzmNlhJNJeerqshVH8dz3Xu/DDlqmQ6NTIdarkat0BAdGkJUSCg6pQaVQoVSrkApV175p6j/qbjy/z7tk9F978Plj6XVGkJZWRnR0dduumIwGDh27NhNn4+OjsZkMjV6ndWrV5OVlfWj5Sq+ZCJzVxYSUpPjFHIFWoWaIJUOnUqHTqUlTBuKMcRAiFpPpC6cCF0YkbpwInXhGIKjcDrh169sweH0AB6UilI0KgUatQKNSolGrSA8WIOxg56wYDUDCjeicVmQaXQotXqUumDkGj0KfSjKkHYoQ6Mu/4uIQRXZHtlP8CSWILQFMpkcjbFzg2/sSV4Pbks57uoyXFUmgipM6Guq6Gapxm21gOUSuO3U4aJG6aVMqWezrA8uuR2b3Ead3AlyOzJFLZX2Ss5fkrC57Li8brySt8k8k1ImMLLbv/3o29lqDcHr9TY4li1JUoPl5p6/avLkyYwbN67BY6WlpY3OR7RUXKiRd0YtwuVxX+m2CpSyy1336rJcJr+l4/BqHfx1wXC8koRapUCpaG638H9uaRsEQfA/mVyBKjwGVXgMuo69CG1mvCRJPHHde57L7cXp8qDTqlDIv/f+KHnrj2BcO3rhxuV1gyTRPiSmVbap1RqC0WgkJyenftlsNmMwGBo8bzab65fLy8sbPH9VaGgooaHNlfpf05rfJb46s6ggCML3Xf8hUyaToVYpUKsa7/nLZXLkCjkqhW9nJG61MxsDBgxg3759VFZWYrPZ2Lp1K4MGXZvHPy4uDo1Gw8GDBwHYsGFDg+cFQRAE32q1hhATE8PMmTOZNGkSY8eOZeTIkfTu3Zv09HRyc3MBWLx4MZmZmQwfPpy6ujomTZrUWnEEQRCEZrTq8Y1Ro0YxalTDCdVWrVpV//+kpCTWrVvXmhEEQRCEFgrML8MKgiAIPicagiAIggCIhiAIgiBc0Sa/I+nxeIAfdsWyIAjCz83V98yr76HXa5MN4er1C7d6cZogCMLPmdlsJiGh8U24ZJIkNT2HQwCy2+3k5eURHR2NQuG76RyuXiG9Zs0ajEajz37vD9VWc4PI7g9tNTeI7M3xeDyYzWZ69uyJVtt4LqQ2uYeg1WpJTU312+83Go3Ex8f77fffqraaG0R2f2iruUFkb8qN9gyuEieVBUEQBEA0BEEQBOEK0RAEQRAEABQLFixY4O8QbYlGoyEtLQ2Npm3dvL6t5gaR3R/aam4Q2X+INvktI0EQBOHHJw4ZCYIgCIBoCIIgCMIVbfI6BF86ePAgmZmZuFwuwsPDee2114iLi2swpqioiJEjR9KhQwcAoqKieO+99/wRt15LcjudTubOnUteXh5arZbFixfTpUsXPyVu7O2330ahUPD88883ei4Qa35VU7kDtebFxcVkZGRQUVFBp06dWLx4MXq9vsGYQKv5xo0bWbFiBW63m8mTJzeauSA/P5+5c+ditVpJTU1l4cKFKJX+f8trLndWVhYff/xx/Z0iH3nkEd/NyiAJTRoyZIiUn58vSZIkrV27Vpo2bVqjMZs3b5bmz5/v62hNaknud999tz53dna29PDDD/s0481YLBZp9uzZUu/evaWlS5fecEwg1rwluQO15lOmTJE2bdokSZIkZWVlSa+//nqjMYFU89LSUmnIkCFSVVWVZLVapVGjRkmnT59uMGbEiBHS4cOHJUmSpNmzZ0tr1qzxR9QGWpJ76tSp0qFDh/ySTxwyaoLT6eSFF14gKSkJgG7dulFSUtJoXG5uLqdOnWLMmDFMmjSJgoICX0dtoKW5v/rqK0aPHg1Av379qKyspLi42KdZb2T79u107NiRp5566qZjAq3m0LLcgVhzl8vFgQMHGDZsGADjx49n8+bNjcYFUs337t1L//79CQ8PJygoiGHDhjXIXFRUhN1uJyUlBbj5Nvlac7kB8vLyWLlyJaNGjeKVV17B4XD4LJ9oCE1Qq9WMGTMGAK/XS1ZWFkOHDm00TqPRMHr0aNavX88zzzzDs88+i9Pp9HXcei3NXVZWRnR0dP1ydHR0QMwgO3bsWKZMmdLkPFWBVnNoWe5ArHlVVRXBwcH1h1Oio6MxmUyNxgVSza+vo8FgaJD5RnW+0Tb5WnO5rVYr3bt3JyMjg/Xr12OxWFi+fLnP8vn/gFqA+Pzzz8nMzGzwWOfOnXn//fdxOp3MmjULt9vN1KlTG637/WPFgwcP5s033+Ts2bP1n9ADNbckSchksgbLcrnvPiM0lb05gVrz5gRizRMSEhpkAhotg39rfj2v19uojt9fbu55f2kul16vb3Cb4aeffpo5c+Ywc+ZMn+QTDeGKBx98kAcffLDR41arlenTpxMeHs6KFStQqVSNxnzwwQeMHDmSiIgI4PIf2Vcnr35I7piYGMrKyupPEpaXl2MwGFo981U3y94SgVjzlgjEmrtcLtLS0vB4PCgUCsxm8w0z+bPm1zMajeTk5NQvX5/ZaDTWT5MPvq/zzTSXu7i4mL179zJhwgTA9zUWh4yakZGRQUJCAm+//TZqtfqGYw4cOMC6desAyM7Oxuv10rlzZ1/GbKQluQcPHsyGDRsAyMnJQaPREBsb68uYtywQa94SgVhzlUpFamoqn332GQCffvopgwYNajQukGo+YMAA9u3bR2VlJTabja1btzbIHBcXh0aj4eDBgwBs2LDhhtvka83l1mq1vPHGG1y8eBFJklizZg3333+/z/KJK5WbcOLECcaNG0fXrl3ru7TBYGDVqlV8+OGHlJWV8cILL2AymZg1axZmsxmNRsOrr77ql93ofzW3w+Hg5ZdfJi8vD7VazaJFi0hOTvZb7ustW7YMuHaoIpBr/n1N5Q7UmhcVFTFr1iwqKipo3749b731FmFhYQFd840bN7Jy5UpcLhcTJkwgPT2d9PR0ZsyYQa9evTh58iTz5s2jtraW5ORkMjMzb/rhKJByb9myhWXLluFyuejbty8LFy70WW7REARBEARAHDISBEEQrhANQRAEQQBEQxAEQRCuEA1BEARBAERDEARBEK4QDUEQBEEAREMQBEEQrhANQRAEQQDg/wGA4tmO4EdhlAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "np.log(mae_no_ll24).plot(kind='density')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gru      (0.8911484479904175, 0.0051367017440497875)\n",
       "cnn     (0.8308230042457581, 0.00025233172345906496)\n",
       "lstm       (0.9328707456588745, 0.05855398625135422)\n",
       "dtype: object"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.log(mae_no_ll24).apply(stats.shapiro)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Se puede apreciar que una transformación logarítmica tampoco soluciona el problema de la no normalidad. Prodecdemos entonces a realizar el test de Friedman."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "FriedmanchisquareResult(statistic=0.46666666666664014, pvalue=0.7918895663367922)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "stats.friedmanchisquare(mae_no_ll24.gru,mae_no_ll24.lstm,mae_no_ll24.cnn)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "En este caso y al 95% de confianza podemos indicar que no existen diferencias significativas (pvalue=0.79>0.05). Esto lo tiene que confirmar el test de Nemenyi."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>gru</th>\n",
       "      <th>cnn</th>\n",
       "      <th>lstm</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>gru</th>\n",
       "      <td>-1.000000</td>\n",
       "      <td>0.848105</td>\n",
       "      <td>0.774254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cnn</th>\n",
       "      <td>0.848105</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>0.900000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lstm</th>\n",
       "      <td>0.774254</td>\n",
       "      <td>0.900000</td>\n",
       "      <td>-1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           gru       cnn      lstm\n",
       "gru  -1.000000  0.848105  0.774254\n",
       "cnn   0.848105 -1.000000  0.900000\n",
       "lstm  0.774254  0.900000 -1.000000"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scp.posthoc_nemenyi_friedman(mae_no_ll24, melted=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Según esto todo serían empates. Pero vamos a comprobarlo con el test de wilcoxon por pares."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>wins</th>\n",
       "      <th>ties</th>\n",
       "      <th>losses</th>\n",
       "      <th>diff</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>gru</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cnn</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lstm</th>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      wins  ties  losses  diff\n",
       "gru    0.0   2.0     0.0   0.0\n",
       "cnn    0.0   2.0     0.0   0.0\n",
       "lstm   0.0   2.0     0.0   0.0"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cols = mae_no_ll24.columns\n",
    "pvalues = np.array([])\n",
    "mae_ranks = pd.DataFrame({\"wins\":np.zeros(3),\n",
    "                     \"ties\":np.zeros(3),\n",
    "                     \"losses\":np.zeros(3)})\n",
    "mae_ranks.index= cols\n",
    "for index1 in range(len(cols)):\n",
    "    for index2 in range(index1+1,len(cols)):\n",
    "        T,p = stats.wilcoxon(mae_no_ll24.iloc[:,index1],mae_no_ll24.iloc[:,index2])\n",
    "        test,padjust,a1,a2 = st.stats.multitest.multipletests(p,alpha = 0.05,method = \"bonferroni\")\n",
    "        if not test:\n",
    "            mae_ranks.iloc[index1,1] = mae_ranks.iloc[index1,1]+1\n",
    "            mae_ranks.iloc[index2,1] = mae_ranks.iloc[index2,1]+1\n",
    "        else:\n",
    "            if mae_no_ll24.iloc[:,index1].mean() > mae_no_ll24.iloc[:,index2].mean():\n",
    "                mae_ranks.iloc[index1,0] = mae_ranks.iloc[index1,0]+1\n",
    "                mae_ranks.iloc[index2,2] = mae_ranks.iloc[index2,2]+1\n",
    "            else:\n",
    "                mae_ranks.iloc[index2,0] = mae_ranks.iloc[index2,0]+1\n",
    "                mae_ranks.iloc[index1,1] = mae_ranks.iloc[index1,2]+1\n",
    "                \n",
    "mae_ranks[\"diff\"] = mae_ranks.wins-mae_ranks.losses\n",
    "mae_ranks = mae_ranks.sort_values(by=\"diff\",ascending = False)\n",
    "mae_ranks"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1c1f112ed0>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "mae_no_ll24.boxplot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test estadísticos paramétricos"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "comprobación de la normalidad"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gru     (0.9473890662193298, 0.14386078715324402)\n",
       "cnn     (0.9143109917640686, 0.01914365030825138)\n",
       "lstm    (0.9627966284751892, 0.36443835496902466)\n",
       "dtype: object"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cc_no_ll24.apply(stats.shapiro)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "En este caso, al 95% de confianza no podemos rechazar la hipótesis nula de que las muestreas de cada modelo procedan de una distribución normal. La otra condición de parametricidad que hay que comprobar el la esfercidad a través del test de Mauchly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(True, 0.9756501456698216, 0.6902339958005032, 2, 0.7081374980205799)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pg.sphericity(cc_no_ll24)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Con el 95% de confianza no podemos rechazar la hipótesis nula de que las muestras cumple la condición de esfericidad (pvalue=0.7 > 0.05). Al cumplirse las dos condiciones podemos aplicar el test ANOVA con medidas repetidas para comprobar si hay diferencias significativas en los métodos. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Source</th>\n",
       "      <th>ddof1</th>\n",
       "      <th>ddof2</th>\n",
       "      <th>F</th>\n",
       "      <th>p-unc</th>\n",
       "      <th>np2</th>\n",
       "      <th>eps</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Within</td>\n",
       "      <td>2</td>\n",
       "      <td>58</td>\n",
       "      <td>16.761015</td>\n",
       "      <td>0.000002</td>\n",
       "      <td>0.366273</td>\n",
       "      <td>0.976229</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Source  ddof1  ddof2          F     p-unc       np2       eps\n",
       "0  Within      2     58  16.761015  0.000002  0.366273  0.976229"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cc_no_ll24.rm_anova()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Como podemos observar, al 95% de confianza, podemos rechazar la hipótesis nula de que no existen diferencias significativas entre los mdelos. Vamos a reslizar un test post-hoc de Tukey para ver dónde están dichas diferencias. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Primero, para poder utilizar la función debemos modificar el data frame para crear una nueva versión utilizando el método `melted()`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>variable</th>\n",
       "      <th>value</th>\n",
       "      <th>subject</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>gru</td>\n",
       "      <td>0.170892</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>gru</td>\n",
       "      <td>0.573044</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>gru</td>\n",
       "      <td>0.406057</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>gru</td>\n",
       "      <td>-0.086384</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>gru</td>\n",
       "      <td>0.146501</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>lstm</td>\n",
       "      <td>0.414315</td>\n",
       "      <td>26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>lstm</td>\n",
       "      <td>0.074001</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>lstm</td>\n",
       "      <td>0.324836</td>\n",
       "      <td>28</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>lstm</td>\n",
       "      <td>0.701757</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>lstm</td>\n",
       "      <td>0.087838</td>\n",
       "      <td>30</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>90 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   variable     value  subject\n",
       "0       gru  0.170892        1\n",
       "1       gru  0.573044        2\n",
       "2       gru  0.406057        3\n",
       "3       gru -0.086384        4\n",
       "4       gru  0.146501        5\n",
       "..      ...       ...      ...\n",
       "85     lstm  0.414315       26\n",
       "86     lstm  0.074001       27\n",
       "87     lstm  0.324836       28\n",
       "88     lstm  0.701757       29\n",
       "89     lstm  0.087838       30\n",
       "\n",
       "[90 rows x 3 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cc_melted = cc_no_ll24.melt()\n",
    "cc_melted[\"subject\"] = [i for _ in range(3) for i in range(1, 31)]\n",
    "cc_melted"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ahora ya podemos realizar el test de Tukey."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>A</th>\n",
       "      <th>B</th>\n",
       "      <th>mean(A)</th>\n",
       "      <th>mean(B)</th>\n",
       "      <th>diff</th>\n",
       "      <th>se</th>\n",
       "      <th>tail</th>\n",
       "      <th>T</th>\n",
       "      <th>p-tukey</th>\n",
       "      <th>hedges</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>cnn</td>\n",
       "      <td>gru</td>\n",
       "      <td>0.356369</td>\n",
       "      <td>0.231382</td>\n",
       "      <td>0.124987</td>\n",
       "      <td>0.068537</td>\n",
       "      <td>two-sided</td>\n",
       "      <td>1.823634</td>\n",
       "      <td>0.163149</td>\n",
       "      <td>0.464745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>cnn</td>\n",
       "      <td>lstm</td>\n",
       "      <td>0.356369</td>\n",
       "      <td>0.261304</td>\n",
       "      <td>0.095065</td>\n",
       "      <td>0.068537</td>\n",
       "      <td>two-sided</td>\n",
       "      <td>1.387051</td>\n",
       "      <td>0.348982</td>\n",
       "      <td>0.353484</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>gru</td>\n",
       "      <td>lstm</td>\n",
       "      <td>0.231382</td>\n",
       "      <td>0.261304</td>\n",
       "      <td>-0.029922</td>\n",
       "      <td>0.068537</td>\n",
       "      <td>two-sided</td>\n",
       "      <td>-0.436583</td>\n",
       "      <td>0.893456</td>\n",
       "      <td>-0.111261</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     A     B   mean(A)   mean(B)      diff        se       tail         T  \\\n",
       "0  cnn   gru  0.356369  0.231382  0.124987  0.068537  two-sided  1.823634   \n",
       "1  cnn  lstm  0.356369  0.261304  0.095065  0.068537  two-sided  1.387051   \n",
       "2  gru  lstm  0.231382  0.261304 -0.029922  0.068537  two-sided -0.436583   \n",
       "\n",
       "    p-tukey    hedges  \n",
       "0  0.163149  0.464745  \n",
       "1  0.348982  0.353484  \n",
       "2  0.893456 -0.111261  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tukey_test = pg.pairwise_tukey(data=cc_melted, dv='value', between='variable')\n",
    "tukey_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}