notebook_6_integrador.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from matplotlib import pyplot\n",
    "from sklearn.decomposition import PCA\n",
    "import pandas as pd\n",
    "\n",
    "np.random.seed(10)"
   ]
  },
  {
   "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>longitud</th>\n",
       "      <th>cant_vocales</th>\n",
       "      <th>cant_consonantes</th>\n",
       "      <th>prop_vocales</th>\n",
       "      <th>prop_consonantes</th>\n",
       "      <th>cant_consonantes_vecinas</th>\n",
       "      <th>cant_vocales_vecinas</th>\n",
       "      <th>cant_consonantes_distintas</th>\n",
       "      <th>cant_vocales_distintas</th>\n",
       "      <th>cant_consonantes_iniciales</th>\n",
       "      <th>...</th>\n",
       "      <th>cant_letra_R</th>\n",
       "      <th>cant_letra_S</th>\n",
       "      <th>cant_letra_T</th>\n",
       "      <th>cant_letra_U</th>\n",
       "      <th>cant_letra_V</th>\n",
       "      <th>cant_letra_W</th>\n",
       "      <th>cant_letra_X</th>\n",
       "      <th>cant_letra_Y</th>\n",
       "      <th>cant_letra_Z</th>\n",
       "      <th>clase</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>acronimo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>0.750000</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>acronimo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>acronimo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>acronimo</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>acronimo</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   longitud  cant_vocales  cant_consonantes  prop_vocales  prop_consonantes  \\\n",
       "0         4             2                 2      0.500000          0.500000   \n",
       "1         4             3                 1      0.250000          0.750000   \n",
       "2         3             2                 1      0.333333          0.666667   \n",
       "3         3             1                 2      0.666667          0.333333   \n",
       "4         3             2                 1      0.333333          0.666667   \n",
       "\n",
       "   cant_consonantes_vecinas  cant_vocales_vecinas  cant_consonantes_distintas  \\\n",
       "0                         1                     1                           2   \n",
       "1                         1                     2                           1   \n",
       "2                         1                     1                           1   \n",
       "3                         1                     1                           2   \n",
       "4                         1                     1                           1   \n",
       "\n",
       "   cant_vocales_distintas  cant_consonantes_iniciales    ...     cant_letra_R  \\\n",
       "0                       1                           1    ...                0   \n",
       "1                       2                           0    ...                0   \n",
       "2                       2                           0    ...                0   \n",
       "3                       1                           1    ...                0   \n",
       "4                       2                           0    ...                0   \n",
       "\n",
       "   cant_letra_S  cant_letra_T  cant_letra_U  cant_letra_V  cant_letra_W  \\\n",
       "0             1             0             0             0             0   \n",
       "1             0             0             0             0             0   \n",
       "2             0             0             1             0             0   \n",
       "3             0             0             0             0             0   \n",
       "4             1             0             0             0             0   \n",
       "\n",
       "   cant_letra_X  cant_letra_Y  cant_letra_Z     clase  \n",
       "0             0             0             0  acronimo  \n",
       "1             0             0             0  acronimo  \n",
       "2             0             0             0  acronimo  \n",
       "3             0             0             0  acronimo  \n",
       "4             0             0             0  acronimo  \n",
       "\n",
       "[5 rows x 40 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "siglas = pd.read_csv(\"siglas-dev.csv\")\n",
    "siglas.head()"
   ]
  },
  {
   "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>pcm_loudness_sma_maxPos</th>\n",
       "      <th>pcm_loudness_sma_minPos</th>\n",
       "      <th>pcm_loudness_sma_amean</th>\n",
       "      <th>pcm_loudness_sma_linregc1</th>\n",
       "      <th>pcm_loudness_sma_linregc2</th>\n",
       "      <th>pcm_loudness_sma_linregerrA</th>\n",
       "      <th>pcm_loudness_sma_linregerrQ</th>\n",
       "      <th>pcm_loudness_sma_stddev</th>\n",
       "      <th>pcm_loudness_sma_skewness</th>\n",
       "      <th>pcm_loudness_sma_kurtosis</th>\n",
       "      <th>...</th>\n",
       "      <th>shimmerLocal_sma_de_quartile3</th>\n",
       "      <th>shimmerLocal_sma_de_iqr1-2</th>\n",
       "      <th>shimmerLocal_sma_de_iqr2-3</th>\n",
       "      <th>shimmerLocal_sma_de_iqr1-3</th>\n",
       "      <th>shimmerLocal_sma_de_percentile99.0</th>\n",
       "      <th>shimmerLocal_sma_de_upleveltime75</th>\n",
       "      <th>shimmerLocal_sma_de_upleveltime90</th>\n",
       "      <th>F0final__Turn_numOnsets</th>\n",
       "      <th>F0final__Turn_duration</th>\n",
       "      <th>gender</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>16</td>\n",
       "      <td>120</td>\n",
       "      <td>0.416503</td>\n",
       "      <td>-0.000172</td>\n",
       "      <td>0.442538</td>\n",
       "      <td>0.171803</td>\n",
       "      <td>0.046154</td>\n",
       "      <td>0.215363</td>\n",
       "      <td>0.648227</td>\n",
       "      <td>3.173967</td>\n",
       "      <td>...</td>\n",
       "      <td>0.019973</td>\n",
       "      <td>0.020172</td>\n",
       "      <td>0.022912</td>\n",
       "      <td>0.043083</td>\n",
       "      <td>0.147333</td>\n",
       "      <td>0.066421</td>\n",
       "      <td>0.018450</td>\n",
       "      <td>14</td>\n",
       "      <td>3.05</td>\n",
       "      <td>m</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>46</td>\n",
       "      <td>4</td>\n",
       "      <td>0.360616</td>\n",
       "      <td>-0.000199</td>\n",
       "      <td>0.389547</td>\n",
       "      <td>0.134897</td>\n",
       "      <td>0.028417</td>\n",
       "      <td>0.169404</td>\n",
       "      <td>0.469004</td>\n",
       "      <td>3.029455</td>\n",
       "      <td>...</td>\n",
       "      <td>0.029609</td>\n",
       "      <td>0.022776</td>\n",
       "      <td>0.033646</td>\n",
       "      <td>0.056422</td>\n",
       "      <td>0.138265</td>\n",
       "      <td>0.046875</td>\n",
       "      <td>0.011719</td>\n",
       "      <td>12</td>\n",
       "      <td>2.94</td>\n",
       "      <td>m</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>179</td>\n",
       "      <td>0</td>\n",
       "      <td>0.423538</td>\n",
       "      <td>-0.000975</td>\n",
       "      <td>0.552752</td>\n",
       "      <td>0.203101</td>\n",
       "      <td>0.057784</td>\n",
       "      <td>0.251776</td>\n",
       "      <td>0.340085</td>\n",
       "      <td>2.080629</td>\n",
       "      <td>...</td>\n",
       "      <td>0.023967</td>\n",
       "      <td>0.023843</td>\n",
       "      <td>0.028746</td>\n",
       "      <td>0.052589</td>\n",
       "      <td>0.133463</td>\n",
       "      <td>0.082927</td>\n",
       "      <td>0.014634</td>\n",
       "      <td>9</td>\n",
       "      <td>2.68</td>\n",
       "      <td>f</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>274</td>\n",
       "      <td>14</td>\n",
       "      <td>0.398092</td>\n",
       "      <td>0.000393</td>\n",
       "      <td>0.339768</td>\n",
       "      <td>0.221290</td>\n",
       "      <td>0.065696</td>\n",
       "      <td>0.258530</td>\n",
       "      <td>0.376392</td>\n",
       "      <td>2.299270</td>\n",
       "      <td>...</td>\n",
       "      <td>0.018540</td>\n",
       "      <td>0.020117</td>\n",
       "      <td>0.020323</td>\n",
       "      <td>0.040440</td>\n",
       "      <td>0.112048</td>\n",
       "      <td>0.063636</td>\n",
       "      <td>0.022727</td>\n",
       "      <td>8</td>\n",
       "      <td>3.00</td>\n",
       "      <td>f</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>194</td>\n",
       "      <td>186</td>\n",
       "      <td>0.412215</td>\n",
       "      <td>0.000181</td>\n",
       "      <td>0.388010</td>\n",
       "      <td>0.187604</td>\n",
       "      <td>0.050548</td>\n",
       "      <td>0.225266</td>\n",
       "      <td>0.445536</td>\n",
       "      <td>2.432539</td>\n",
       "      <td>...</td>\n",
       "      <td>0.015643</td>\n",
       "      <td>0.014819</td>\n",
       "      <td>0.019811</td>\n",
       "      <td>0.034629</td>\n",
       "      <td>0.102689</td>\n",
       "      <td>0.082192</td>\n",
       "      <td>0.018265</td>\n",
       "      <td>10</td>\n",
       "      <td>2.71</td>\n",
       "      <td>f</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1583 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   pcm_loudness_sma_maxPos  pcm_loudness_sma_minPos  pcm_loudness_sma_amean  \\\n",
       "0                       16                      120                0.416503   \n",
       "1                       46                        4                0.360616   \n",
       "2                      179                        0                0.423538   \n",
       "3                      274                       14                0.398092   \n",
       "4                      194                      186                0.412215   \n",
       "\n",
       "   pcm_loudness_sma_linregc1  pcm_loudness_sma_linregc2  \\\n",
       "0                  -0.000172                   0.442538   \n",
       "1                  -0.000199                   0.389547   \n",
       "2                  -0.000975                   0.552752   \n",
       "3                   0.000393                   0.339768   \n",
       "4                   0.000181                   0.388010   \n",
       "\n",
       "   pcm_loudness_sma_linregerrA  pcm_loudness_sma_linregerrQ  \\\n",
       "0                     0.171803                     0.046154   \n",
       "1                     0.134897                     0.028417   \n",
       "2                     0.203101                     0.057784   \n",
       "3                     0.221290                     0.065696   \n",
       "4                     0.187604                     0.050548   \n",
       "\n",
       "   pcm_loudness_sma_stddev  pcm_loudness_sma_skewness  \\\n",
       "0                 0.215363                   0.648227   \n",
       "1                 0.169404                   0.469004   \n",
       "2                 0.251776                   0.340085   \n",
       "3                 0.258530                   0.376392   \n",
       "4                 0.225266                   0.445536   \n",
       "\n",
       "   pcm_loudness_sma_kurtosis   ...    shimmerLocal_sma_de_quartile3  \\\n",
       "0                   3.173967   ...                         0.019973   \n",
       "1                   3.029455   ...                         0.029609   \n",
       "2                   2.080629   ...                         0.023967   \n",
       "3                   2.299270   ...                         0.018540   \n",
       "4                   2.432539   ...                         0.015643   \n",
       "\n",
       "   shimmerLocal_sma_de_iqr1-2  shimmerLocal_sma_de_iqr2-3  \\\n",
       "0                    0.020172                    0.022912   \n",
       "1                    0.022776                    0.033646   \n",
       "2                    0.023843                    0.028746   \n",
       "3                    0.020117                    0.020323   \n",
       "4                    0.014819                    0.019811   \n",
       "\n",
       "   shimmerLocal_sma_de_iqr1-3  shimmerLocal_sma_de_percentile99.0  \\\n",
       "0                    0.043083                            0.147333   \n",
       "1                    0.056422                            0.138265   \n",
       "2                    0.052589                            0.133463   \n",
       "3                    0.040440                            0.112048   \n",
       "4                    0.034629                            0.102689   \n",
       "\n",
       "   shimmerLocal_sma_de_upleveltime75  shimmerLocal_sma_de_upleveltime90  \\\n",
       "0                           0.066421                           0.018450   \n",
       "1                           0.046875                           0.011719   \n",
       "2                           0.082927                           0.014634   \n",
       "3                           0.063636                           0.022727   \n",
       "4                           0.082192                           0.018265   \n",
       "\n",
       "   F0final__Turn_numOnsets  F0final__Turn_duration  gender  \n",
       "0                       14                    3.05       m  \n",
       "1                       12                    2.94       m  \n",
       "2                        9                    2.68       f  \n",
       "3                        8                    3.00       f  \n",
       "4                       10                    2.71       f  \n",
       "\n",
       "[5 rows x 1583 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "genero = pd.read_csv(\"genero.csv\")\n",
    "display(genero.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ejercicio\n",
    "Experimentar con diferentes algoritmos de clasificación (árboles, Naive Bayes, SVM, KNN) para los ejemplos A\n",
    "y B, usando 10-fold CV. ¿Qué algoritmos funcionan mejor para los datos originales?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ejercicio\n",
    "Rankear los atributos de cada ejemplo según su ganancia de información. Conservar sólo los primeros $k$\n",
    "atributos (probar con distintos valores de $k$) y ver cómo afecta esto al desempeño de los algoritmos del ejercicio\n",
    "anterior. ¿Hay diferencia en los atributos encontrados si la seleccion se hace al principio o en cada iteracion del folding?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ejercicio\n",
    "Experimentar con otros métodos de selección de atributos.\n",
    "Utilizar las siguientes políticas de selección de atributos y comparar cómo funcionan en los dos datasets\n",
    "\n",
    "- Ranking de atributos por p-valor: Para cada atributo comparar usando un test de hipótesis  [ttest](http://docs.scipy.org/doc/scipy-0.14.0/reference/generated/scipy.stats.ttest_ind.html), [kruskall](http://docs.scipy.org/doc/scipy-0.14.0/reference/generated/scipy.stats.mstats.kruskalwallis.html), [ks_test](http://docs.scipy.org/doc/scipy-0.15.1/reference/generated/scipy.stats.ks_2samp.html), etc. la diferencia entre la proyeccion del atributo separando por clase y se quede con los $k$  más significativamente distintos.\n",
    "- Random Search: Elegir un subset de $k$ atributos al azar\n",
    "- [Genetic Search](http://topepo.github.io/caret/feature-selection-using-genetic-algorithms.html)\n",
    "- Recursive Feature Elimination utilizando Random Forest como algoritmo base.\n",
    "\n",
    "Usando los distintosalgoritmos de clasificación vistos en clase, ¿cómo afecta al desempeño hacer selección de atributos?\n",
    "En general, ¿cuán sensible es cada algoritmo de aprendizaje a la cantidad y calidad de atributos?\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Genero n puntos (x,y) a distancia e de la recta y=m*x+b,\n",
    "# donde e=rand[0,1]*d.\n",
    "def genDatos(n, b, m, d):\n",
    "    tmp = []\n",
    "    alfa = np.arctan(m)\n",
    "    for x in range(n):\n",
    "        y = m * x + b\n",
    "        e = np.random.uniform(0.0, 1.0) * d\n",
    "        x -= e * np.sin(alfa)\n",
    "        y += e * np.cos(alfa)\n",
    "        tmp.append([x,y])\n",
    "    return np.array(tmp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD8CAYAAAB5Pm/hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAr2IyMAp0IuIDJwCvYjIwCnQi4gM3P8Hpt+kvkqTXewAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Genero dos muestras de puntos y las ploteo.\n",
    "rojo = genDatos(100, 0, 1, 10)\n",
    "azul = genDatos(100, 0, 1, -10)\n",
    "pyplot.plot(rojo[:,0], rojo[:,1], 'ro')\n",
    "pyplot.plot(azul[:,0], azul[:,1], 'bs')\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PCA(copy=True, iterated_power='auto', n_components=2, random_state=None,\n",
       "  svd_solver='auto', tol=0.0, whiten=False)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Computo PCA sobre todos los puntos.\n",
    "# http://scikit-learn.org/stable/modules/generated/sklearn.decomposition.PCA.html\n",
    "todos = np.concatenate((rojo, azul))\n",
    "pca = PCA(n_components=2)\n",
    "pca.fit(todos)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD8CAYAAABzTgP2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niIiMUWAQEZExCgwiIjJGgUFERMb8f/Wh6c4qTgtEAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Transformo las muestras originales al nuevo sistema de coordenadas,\n",
    "# y las ploteo.\n",
    "rojo_transf = pca.transform(rojo)\n",
    "azul_transf = pca.transform(azul)\n",
    "pyplot.plot(rojo_transf[:,0], rojo_transf[:,1], 'ro')\n",
    "pyplot.plot(azul_transf[:,0], azul_transf[:,1], 'bs')\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ejercicio \n",
    "Transformar con PCA los datos de los ejemplos A y B, conservar sólo las primeras $k$\n",
    "Componentes (probar con distintos valores de $k$) y volver a evaluar los algoritmos de clasificación."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}