KNN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "283dca66-01a4-409e-8b28-a865f8c93507",
   "metadata": {},
   "source": [
    "Bernat Sort Rufat\n",
    "\n",
    "MD005 Sistemas Basados en el Conocimiento"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c64fd068-3d45-4c43-8f54-b6a7fce51113",
   "metadata": {},
   "source": [
    "# Ejercicio 3.4: Aprendizaje Analógico - KNN (Python)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4f5ac622-0596-49f3-880e-b8c8473e0083",
   "metadata": {},
   "source": [
    "## Librerías "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "2f4028b6-c039-477b-a7b1-86422e9f0803",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.io import arff\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "\n",
    "from imblearn.over_sampling import SMOTE\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import sklearn.metrics as metrics\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.metrics import recall_score\n",
    "from sklearn.metrics import classification_report"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3041079-40e6-402c-afeb-873e899b5009",
   "metadata": {},
   "source": [
    "## glass.arff"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "cfc1e089-d09f-47b0-ab87-dcdb9e652837",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# cargamos el .arff file\n",
    "glass = arff.loadarff('glass.arff')\n",
    "# glass es una tupla que contiene dos elementos: un array numpy y una lista de tuplas\n",
    "# primer elemento de la tupla (el array numpy) es el que nos interesa (glass[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c73f85d2-4a2c-437d-970d-778f692df3f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "# convertimos los datos a df\n",
    "glass_df = pd.DataFrame(glass[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e826e76e-1cf6-4235-afbb-19fe952bf285",
   "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>RI</th>\n",
       "      <th>Na</th>\n",
       "      <th>Mg</th>\n",
       "      <th>Al</th>\n",
       "      <th>Si</th>\n",
       "      <th>'K'</th>\n",
       "      <th>Ca</th>\n",
       "      <th>Ba</th>\n",
       "      <th>Fe</th>\n",
       "      <th>Type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.51793</td>\n",
       "      <td>12.79</td>\n",
       "      <td>3.50</td>\n",
       "      <td>1.12</td>\n",
       "      <td>73.03</td>\n",
       "      <td>0.64</td>\n",
       "      <td>8.77</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>b'build wind float'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.51643</td>\n",
       "      <td>12.16</td>\n",
       "      <td>3.52</td>\n",
       "      <td>1.35</td>\n",
       "      <td>72.89</td>\n",
       "      <td>0.57</td>\n",
       "      <td>8.53</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>b'vehic wind float'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.51793</td>\n",
       "      <td>13.21</td>\n",
       "      <td>3.48</td>\n",
       "      <td>1.41</td>\n",
       "      <td>72.64</td>\n",
       "      <td>0.59</td>\n",
       "      <td>8.43</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>b'build wind float'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.51299</td>\n",
       "      <td>14.40</td>\n",
       "      <td>1.74</td>\n",
       "      <td>1.54</td>\n",
       "      <td>74.55</td>\n",
       "      <td>0.00</td>\n",
       "      <td>7.59</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>b'tableware'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.53393</td>\n",
       "      <td>12.30</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>70.16</td>\n",
       "      <td>0.12</td>\n",
       "      <td>16.19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.24</td>\n",
       "      <td>b'build wind non-float'</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        RI     Na    Mg    Al     Si   'K'     Ca   Ba    Fe  \\\n",
       "0  1.51793  12.79  3.50  1.12  73.03  0.64   8.77  0.0  0.00   \n",
       "1  1.51643  12.16  3.52  1.35  72.89  0.57   8.53  0.0  0.00   \n",
       "2  1.51793  13.21  3.48  1.41  72.64  0.59   8.43  0.0  0.00   \n",
       "3  1.51299  14.40  1.74  1.54  74.55  0.00   7.59  0.0  0.00   \n",
       "4  1.53393  12.30  0.00  1.00  70.16  0.12  16.19  0.0  0.24   \n",
       "\n",
       "                      Type  \n",
       "0      b'build wind float'  \n",
       "1      b'vehic wind float'  \n",
       "2      b'build wind float'  \n",
       "3             b'tableware'  \n",
       "4  b'build wind non-float'  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 5 primeras filas del df\n",
    "glass_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "1a48f3ec-385d-43d2-83c0-30dbadfb3e4b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([b'build wind float', b'vehic wind float', b'tableware',\n",
       "       b'build wind non-float', b'headlamps', b'containers'], dtype=object)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# nombres de las clases\n",
    "glass_df['Type'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "32277484-44fa-49be-8006-2dabd2087121",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['build_wind_float', 'vehic_wind_float', 'tableware',\n",
       "       'build_wind_non_float', 'headlamps', 'containers'], dtype=object)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# renombramos las clases para un mejor entendimiento \n",
    "glass_df['Type'].replace({b'build wind float': 'build_wind_float',\n",
    "                          b'vehic wind float': 'vehic_wind_float',\n",
    "                          b'tableware': 'tableware',\n",
    "                          b'build wind non-float': 'build_wind_non_float',\n",
    "                          b'headlamps': 'headlamps',\n",
    "                          b'containers': 'containers'}, inplace = True)\n",
    "glass_df['Type'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "7dd46f50-084a-4190-af76-dc4c40054c0e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# renombramos la columna 'K' por K\n",
    "glass_df.rename(columns={\"'K'\": 'K'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "2355ef83-5270-4ef7-80d0-a7fcafb6941e",
   "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>RI</th>\n",
       "      <th>Na</th>\n",
       "      <th>Mg</th>\n",
       "      <th>Al</th>\n",
       "      <th>Si</th>\n",
       "      <th>K</th>\n",
       "      <th>Ca</th>\n",
       "      <th>Ba</th>\n",
       "      <th>Fe</th>\n",
       "      <th>Type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.51793</td>\n",
       "      <td>12.79</td>\n",
       "      <td>3.50</td>\n",
       "      <td>1.12</td>\n",
       "      <td>73.03</td>\n",
       "      <td>0.64</td>\n",
       "      <td>8.77</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>build_wind_float</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.51643</td>\n",
       "      <td>12.16</td>\n",
       "      <td>3.52</td>\n",
       "      <td>1.35</td>\n",
       "      <td>72.89</td>\n",
       "      <td>0.57</td>\n",
       "      <td>8.53</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>vehic_wind_float</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.51793</td>\n",
       "      <td>13.21</td>\n",
       "      <td>3.48</td>\n",
       "      <td>1.41</td>\n",
       "      <td>72.64</td>\n",
       "      <td>0.59</td>\n",
       "      <td>8.43</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>build_wind_float</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.51299</td>\n",
       "      <td>14.40</td>\n",
       "      <td>1.74</td>\n",
       "      <td>1.54</td>\n",
       "      <td>74.55</td>\n",
       "      <td>0.00</td>\n",
       "      <td>7.59</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>tableware</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.53393</td>\n",
       "      <td>12.30</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>70.16</td>\n",
       "      <td>0.12</td>\n",
       "      <td>16.19</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.24</td>\n",
       "      <td>build_wind_non_float</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        RI     Na    Mg    Al     Si     K     Ca   Ba    Fe  \\\n",
       "0  1.51793  12.79  3.50  1.12  73.03  0.64   8.77  0.0  0.00   \n",
       "1  1.51643  12.16  3.52  1.35  72.89  0.57   8.53  0.0  0.00   \n",
       "2  1.51793  13.21  3.48  1.41  72.64  0.59   8.43  0.0  0.00   \n",
       "3  1.51299  14.40  1.74  1.54  74.55  0.00   7.59  0.0  0.00   \n",
       "4  1.53393  12.30  0.00  1.00  70.16  0.12  16.19  0.0  0.24   \n",
       "\n",
       "                   Type  \n",
       "0      build_wind_float  \n",
       "1      vehic_wind_float  \n",
       "2      build_wind_float  \n",
       "3             tableware  \n",
       "4  build_wind_non_float  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 5 primeras filas del df correctox\n",
    "glass_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "b8d9f7ba-0150-4747-b5ce-cccb0cc6556a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 214 entries, 0 to 213\n",
      "Data columns (total 10 columns):\n",
      " #   Column  Non-Null Count  Dtype  \n",
      "---  ------  --------------  -----  \n",
      " 0   RI      214 non-null    float64\n",
      " 1   Na      214 non-null    float64\n",
      " 2   Mg      214 non-null    float64\n",
      " 3   Al      214 non-null    float64\n",
      " 4   Si      214 non-null    float64\n",
      " 5   K       214 non-null    float64\n",
      " 6   Ca      214 non-null    float64\n",
      " 7   Ba      214 non-null    float64\n",
      " 8   Fe      214 non-null    float64\n",
      " 9   Type    214 non-null    object \n",
      "dtypes: float64(9), object(1)\n",
      "memory usage: 16.8+ KB\n"
     ]
    }
   ],
   "source": [
    "# info de las columnas del df\n",
    "glass_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0512561-5656-4f98-8907-f511e9e191e9",
   "metadata": {},
   "source": [
    "- Observamos que el df glass_df tiene 214 filas y 10 columnas. \n",
    "- Los tipos de datos de las columnas son float64 para 9 de las columnas y object para 1 columna. \n",
    "- Los nombres de las columnas son \"RI\", \"Na\", \"Mg\", \"Al\", \"Si\", \"K\", \"Ca\", \"Ba\", \"Fe\" y \"Type\".\n",
    "- Todas las columnas tienen 214 valores no nulos, lo que significa que no hay valores perdidos (missing values) en el df.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "cf4e7a7d-4109-48bf-b01e-412b4a7c264b",
   "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>RI</th>\n",
       "      <th>Na</th>\n",
       "      <th>Mg</th>\n",
       "      <th>Al</th>\n",
       "      <th>Si</th>\n",
       "      <th>K</th>\n",
       "      <th>Ca</th>\n",
       "      <th>Ba</th>\n",
       "      <th>Fe</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "      <td>214.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.518365</td>\n",
       "      <td>13.407850</td>\n",
       "      <td>2.684533</td>\n",
       "      <td>1.444907</td>\n",
       "      <td>72.650935</td>\n",
       "      <td>0.497056</td>\n",
       "      <td>8.956963</td>\n",
       "      <td>0.175047</td>\n",
       "      <td>0.057009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.003037</td>\n",
       "      <td>0.816604</td>\n",
       "      <td>1.442408</td>\n",
       "      <td>0.499270</td>\n",
       "      <td>0.774546</td>\n",
       "      <td>0.652192</td>\n",
       "      <td>1.423153</td>\n",
       "      <td>0.497219</td>\n",
       "      <td>0.097439</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.511150</td>\n",
       "      <td>10.730000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.290000</td>\n",
       "      <td>69.810000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>5.430000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>1.516522</td>\n",
       "      <td>12.907500</td>\n",
       "      <td>2.115000</td>\n",
       "      <td>1.190000</td>\n",
       "      <td>72.280000</td>\n",
       "      <td>0.122500</td>\n",
       "      <td>8.240000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.517680</td>\n",
       "      <td>13.300000</td>\n",
       "      <td>3.480000</td>\n",
       "      <td>1.360000</td>\n",
       "      <td>72.790000</td>\n",
       "      <td>0.555000</td>\n",
       "      <td>8.600000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.519157</td>\n",
       "      <td>13.825000</td>\n",
       "      <td>3.600000</td>\n",
       "      <td>1.630000</td>\n",
       "      <td>73.087500</td>\n",
       "      <td>0.610000</td>\n",
       "      <td>9.172500</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.533930</td>\n",
       "      <td>17.380000</td>\n",
       "      <td>4.490000</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>75.410000</td>\n",
       "      <td>6.210000</td>\n",
       "      <td>16.190000</td>\n",
       "      <td>3.150000</td>\n",
       "      <td>0.510000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               RI          Na          Mg          Al          Si           K  \\\n",
       "count  214.000000  214.000000  214.000000  214.000000  214.000000  214.000000   \n",
       "mean     1.518365   13.407850    2.684533    1.444907   72.650935    0.497056   \n",
       "std      0.003037    0.816604    1.442408    0.499270    0.774546    0.652192   \n",
       "min      1.511150   10.730000    0.000000    0.290000   69.810000    0.000000   \n",
       "25%      1.516522   12.907500    2.115000    1.190000   72.280000    0.122500   \n",
       "50%      1.517680   13.300000    3.480000    1.360000   72.790000    0.555000   \n",
       "75%      1.519157   13.825000    3.600000    1.630000   73.087500    0.610000   \n",
       "max      1.533930   17.380000    4.490000    3.500000   75.410000    6.210000   \n",
       "\n",
       "               Ca          Ba          Fe  \n",
       "count  214.000000  214.000000  214.000000  \n",
       "mean     8.956963    0.175047    0.057009  \n",
       "std      1.423153    0.497219    0.097439  \n",
       "min      5.430000    0.000000    0.000000  \n",
       "25%      8.240000    0.000000    0.000000  \n",
       "50%      8.600000    0.000000    0.000000  \n",
       "75%      9.172500    0.000000    0.100000  \n",
       "max     16.190000    3.150000    0.510000  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# estadísticas descriptivas \n",
    "glass_df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "962a25ad-38cc-4720-845f-244f0d4a6844",
   "metadata": {},
   "source": [
    "El método df.describe() genera estadísticas descriptivas de un df.\n",
    "\n",
    "Obtenemos el recuento, la media, la desviación estándar, el mínimo, el primer cuartil (25%), la mediana (50%), el tercer cuartil (75%) y el máximo de cada columna.\n",
    "\n",
    "Esta información puede ser útil para comprender la distribución y el rango de valores de cada columna del df.\n",
    "\n",
    "Podemos observar, por ejemplo, que para la columna Ba, el valor mínimo es 0, lo que significa que al menos algunas de las muestras no contienen bario. Los valores del primer cuartil (25%), de la mediana (50%) y del tercer cuartil (75%) también son 0, lo que sugiere que la mayoría de las muestras contienen muy poco o nada de bario. El valor máximo es 3.15, que es relativamente pequeño en comparación con los demás componentes químicos presentes en el vidrio.\n",
    "\n",
    "Para la columna Fe, el valor mínimo es 0, lo que significa que al menos algunas muestras no contienen hierro. El primer cuartil (25%), la mediana (50%), el tercer cuartil (75%) y el valor máximo son 0, 0, 0.1 y 0.51 respectivamente. Esto sugiere que la mayoría de las muestras contienen muy poco o nada de hierro.\n",
    "\n",
    "Parece ser que el bario y el hierro son componentes relativamente raros en las muestras de vidrio, ya que la mayoría de las muestras contienen poca cantidad. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd06989e-c421-4ff4-a1c1-78ff8e958bbf",
   "metadata": {},
   "source": [
    "### Distribución de cada variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "910e880f-8b82-4c75-9c77-d9450d1558b1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2500x1500 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# creating a figure and a 3x3 grid of subplots\n",
    "fig, axs = plt.subplots(nrows=3, ncols=3, figsize=(25, 15))\n",
    "\n",
    "# seaborn's histplot to plot histograms on the subplots\n",
    "sns.histplot(data=glass_df, x='RI', element='step', stat='density', ax=axs[0, 0])\n",
    "sns.histplot(data=glass_df, x='Na', element='step', stat='density', ax=axs[0, 1])\n",
    "sns.histplot(data=glass_df, x='Mg', element='step', stat='density', ax=axs[0, 2])\n",
    "sns.histplot(data=glass_df, x='Al', element='step', stat='density', ax=axs[1, 0])\n",
    "sns.histplot(data=glass_df, x='Si', element='step', stat='density', ax=axs[1, 1])\n",
    "sns.histplot(data=glass_df, x='K', element='step', stat='density', ax=axs[1, 2])\n",
    "sns.histplot(data=glass_df, x='Ca', element='step', stat='density', ax=axs[2, 0])\n",
    "sns.histplot(data=glass_df, x='Ba', element='step', stat='density', ax=axs[2, 1])\n",
    "sns.histplot(data=glass_df, x='Fe', element='step', stat='density', ax=axs[2, 2])\n",
    "\n",
    "# axis labels\n",
    "axs[0, 0].set(xlabel='RI', ylabel='Density')\n",
    "axs[0, 1].set(xlabel='Na', ylabel='Density')\n",
    "axs[0, 2].set(xlabel='Mg', ylabel='Density')\n",
    "axs[1, 0].set(xlabel='Al', ylabel='Density')\n",
    "axs[1, 1].set(xlabel='Si', ylabel='Density')\n",
    "axs[1, 2].set(xlabel='K', ylabel='Density')\n",
    "axs[2, 0].set(xlabel='Ca', ylabel='Density')\n",
    "axs[2, 1].set(xlabel='Ba', ylabel='Density')\n",
    "axs[2, 2].set(xlabel='Fe', ylabel='Density')\n",
    "\n",
    "# adjusting the spacing between the subplots\n",
    "plt.tight_layout()\n",
    "\n",
    "# showing the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e3dc20c-c8b3-4484-a66f-6efbd1e1c740",
   "metadata": {},
   "source": [
    "- Observamos que algunas variables se distribuyen de manera normal (como RI, Na, Ca, Al y Si) mientras que otras presentan una asimetría positiva (por ejemplo, Ba, Fe y K) o negativa (por ejemplo, Mg). "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4400c6d8-1a80-45e9-87d6-eb231191df29",
   "metadata": {},
   "source": [
    "### Comprobación del desequilibrio de clases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "b8072f34-1b81-4e1b-9120-65f562815614",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "build_wind_non_float    76\n",
      "build_wind_float        70\n",
      "headlamps               29\n",
      "vehic_wind_float        17\n",
      "containers              13\n",
      "tableware                9\n",
      "Name: Type, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Checking data imbalance \n",
    "print(glass_df['Type'].value_counts()) # recuento de instancias de cada clase en el df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5a76522a-9ffc-4675-a273-84bc5edccf07",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1600x800 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(16,8))\n",
    "sns.countplot(x = glass_df['Type'], color = 'blue')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bf101c8a-dcea-4423-9ec7-2f528a13876b",
   "metadata": {},
   "source": [
    "- Podemos observar que hay un desequilibrio de clases.\n",
    "- Por ejemplo, vemos que la clase build_wind_non_float tiene 76 instancias, mientras que la clase tableware solo tiene 9 instancias. Esto significa que la clase build_wind_non_float tiene más de 8 veces el número de instancias que la clase tableware.\n",
    "- Este desequilibrio en el número de instancias de cada clase puede plantear problemas a la hora de entrenar y evaluar modelos de Machine Learning. Por ejemplo, un modelo que siempre predice la clase mayoritaria es probable que logre una alta accuracy, pero no será útil en la práctica, ya que no está generalizando bien."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03f50854-8152-41f1-8cfa-237a6ffa2c06",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Separar las variables independientes de la variable dependiente\n",
    "- Separar en atributos (X) y etiquetas (y).\n",
    "- Las columnas del conjunto de datos se pueden separar en feature columns (X) y target column (y)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "3db88349-b4e4-4083-b72f-0b4bd04921ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "# variables independientes (feature columns): las variables que se utilizan para predecir la variable objetivo (y)\n",
    "X = glass_df.iloc[:, :-1].values\n",
    "# variable dependiente (target column): en nuestro caso 'Type'\n",
    "y = glass_df.iloc[:, -1].values"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "219b2e77-f779-48fd-a3e6-e260b613cef7",
   "metadata": {},
   "source": [
    "### Oversampling con SMOTE "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f21b7404-de9c-4c1d-955c-f7107d8ae1b1",
   "metadata": {},
   "source": [
    "- Al tratarse de datos desequilibrados, nuestro modelo podría intentar ajustarse a la clase mayoritaria y proporcionar una predicción sesgada, además de ofrecer una accuracy engañosa.\n",
    "- Por este motivo, realizaremos un procedimiento de sobremuestreo en las clases minoritarias con SMOTE."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8f200b85-0916-4946-b58a-df68fd5db687",
   "metadata": {},
   "outputs": [],
   "source": [
    "SMOTE = SMOTE()\n",
    "X_o, y_o = SMOTE.fit_resample(X,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "05648163-f089-4cc8-88c6-5f88b7d11135",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "build_wind_float        76\n",
       "build_wind_non_float    76\n",
       "containers              76\n",
       "headlamps               76\n",
       "tableware               76\n",
       "vehic_wind_float        76\n",
       "dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# miramos el resultado de sobremuestro\n",
    "y1 = pd.DataFrame(y_o)\n",
    "y1.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "4642a836-fcaa-4904-b433-0b0121954766",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2000x500 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "figure, axes = plt.subplots(1, 2, sharex=True,figsize=(20, 5))\n",
    "figure.suptitle('over sampling')\n",
    "axes[0].set_title('antes del oversampling')\n",
    "axes[1].set_title('después del oversampling')\n",
    "sns.countplot(x = y, color = 'blue',ax=axes[0])\n",
    "sns.countplot(x = y_o, color = 'blue',ax=axes[1])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e942d8ec-663d-4e06-81bf-41a67c3bb088",
   "metadata": {},
   "source": [
    "- Hemos incrementa de forma “sintética” los ejemplos de la clase minoritaria.\n",
    "- Podemos observar que ya no hay un desequilibrio de clases, las clases están balanceadas."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e967d813-8b49-4f20-8293-9132bb7bfcb7",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Escalar los atributos"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1eeb4f4a-733b-49c6-beef-61b85294bc73",
   "metadata": {},
   "source": [
    "- Escalamos los datos después de aplicar SMOTE ya que después de sobremuestrear los datos, la varianza de los datos aumentará artificialmente, por lo que el escalado puede ayudar a solucionar este problema. Además, tras el sobremuestreo, la distancia entre las features de la muestra sintética se acercará más a los datos originales, lo que puede ser beneficioso para algoritmos basados en la distancia como el KNN."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "59e0b597-8510-497b-b4e5-4f9d6b10b86b",
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler()\n",
    "scaler.fit(X_o)\n",
    "\n",
    "X_scaled = scaler.transform(X_o)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2dce18d5-55d2-48e1-bbf7-b299479a8b63",
   "metadata": {},
   "source": [
    "- Escalamos para que el algoritmo le dé la misma importancia a todos los datos.\n",
    "- Algoritmos como KNN se basan en la distancia entre los puntos de datos, por lo que es importante que las features estén en la misma escala.\n",
    "- El resultado es un nuevo conjunto de valores que tienen una distribución normal estándar (una distribución gaussiana con una media de 0 y una desviación estándar de 1).\n",
    "- Ahora los tenemos todos dentro de un mismo rango o intervalo.\n",
    "- La variable X_scaled contiene la versión estandarizada de la feature data almacenada en X_o."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ce2aeb0f-4763-4483-a234-e1cf78b85137",
   "metadata": {},
   "source": [
    "### Algoritmo KNN"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2d5a9b30-b58a-425b-b2c7-40b43b79dc9e",
   "metadata": {},
   "source": [
    "#### Para k=2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d635fe04-ef39-44e8-b5a2-e9e96c851725",
   "metadata": {
    "tags": []
   },
   "source": [
    "##### Preparación algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "9768a316-86f1-4ee3-8678-252f933d14b4",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "knn_2 = KNeighborsClassifier(n_neighbors = 2)\n",
    "\n",
    "# 10-Fold Stratified Cross-Validation \n",
    "skf = StratifiedKFold(n_splits=10, random_state=None, shuffle=False)\n",
    "# random_state = None: el generador de números aleatorios utilizará un valor de semilla diferente \n",
    "    # cada vez que se ejecute el código, lo que resultará en un barajado diferente de los datos cada vez.\n",
    "# shuffle=False: los datos se utilizan en el orden del conjunto de datos de entrada y no se barajan.\n",
    "\n",
    "\n",
    "## Arrays para almacenar los resultados de cada iteración\n",
    "cm_list = []\n",
    "accu_list = []\n",
    "prec_list = []\n",
    "recall_list = []\n",
    "f1_list = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f1178c37-96e3-484c-9b71-6c9dd8c52f5e",
   "metadata": {},
   "source": [
    "- Cuando random_state se establece en None y el parámetro shuffle es False, los datos siempre se dividen tal cual, sin barajar ni aleatorizar los datos. Esto significa que las observaciones se utilizan en el orden del conjunto de datos de entrada.\n",
    "- Por tanto, en este caso, cuando random_state=None y shuffle=False, significa que no hay aleatorización al crear los pliegues (folds), los datos no se barajan y las observaciones se utilizan en el orden del conjunto de datos de entrada. El resultado será el mismo cada vez que se ejecute el código.\n",
    "- Esto puede ser útil queremos comparar los resultados del modelo bajo diferentes condiciones, o si queremos asegurar la reproducibilidad de los resultados (queremos que utilice la misma división cada vez que se ejecute el código).\n",
    "- No obstante, dejando de lado el ejercicio, es bueno que vaya cambiando, ya que si vemos que el cojunto de training y test varían mucho, no tenemos datos estables para el problema que queremos resolver."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f4d28191-8dda-4fb9-b2bf-37dbebcf4281",
   "metadata": {},
   "source": [
    "##### Ejecución algoritmo KNN"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b39053de-8c5f-474e-8065-f92b79bede9f",
   "metadata": {},
   "source": [
    "- Se trata de un problema de clasificación multiclase.\n",
    "- average='macro' calculará la métrica independientemente para cada clase y luego sacará la media. Las métricas macro promediadas darán el mismo peso a cada clase, independientemente del desequilibrio de clases.\n",
    "    - En caso de sobremuestreo, el número de muestras en las diferentes clases no está equilibrado, por lo que el uso de micro se verá afectado por este desequilibrio. Macro, en cambio, dará el mismo peso a todas las clases, por lo que es más apropiado. En métodos de sobremuestreo como SMOTE, esto puede ayudar a mitigar el sesgo hacia la clase sobremuestreada y dar una visión más representativa del rendimiento del modelo. Por lo tanto, cuando se utilizan técnicas de sobremuestreo, se recomienda utilizar la opción average='macro' para calcular la *precision*, el *recall* y el *f1-score*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "72f39e0d-07a5-4441-9287-1025511958dd",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteración: 1\n",
      "Fold accuracy: 0.89\n",
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 2]\n",
      " [2 6 0 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 0 6 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.8913043478260869\n",
      "Precision: 0.911111111111111\n",
      "Recall: 0.8928571428571429\n",
      "F1-Score: 0.8958318370083077\n",
      "******\n",
      "\n",
      "Iteración: 2\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 1]\n",
      " [1 6 1 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [1 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.9130434782608695\n",
      "Precision: 0.923611111111111\n",
      "Recall: 0.9166666666666666\n",
      "F1-Score: 0.916141456582633\n",
      "******\n",
      "\n",
      "Iteración: 3\n",
      "Fold accuracy: 0.83\n",
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 2]\n",
      " [3 5 0 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 1 5 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [1 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.8260869565217391\n",
      "Precision: 0.8640572390572391\n",
      "Recall: 0.8244047619047619\n",
      "F1-Score: 0.8292199200867931\n",
      "******\n",
      "\n",
      "Iteración: 4\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[8 0 0 0 0 0]\n",
      " [1 6 0 0 0 1]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [2 0 0 0 0 5]]\n",
      "\n",
      "Accuracy: 0.9130434782608695\n",
      "Precision: 0.9267676767676768\n",
      "Recall: 0.9107142857142857\n",
      "F1-Score: 0.9114131482552534\n",
      "******\n",
      "\n",
      "Iteración: 5\n",
      "Fold accuracy: 0.89\n",
      "Matriz de Confusión:\n",
      " [[7 1 0 0 0 0]\n",
      " [2 4 1 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 1 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.8913043478260869\n",
      "Precision: 0.8888888888888888\n",
      "Recall: 0.8839285714285713\n",
      "F1-Score: 0.8838612368024134\n",
      "******\n",
      "\n",
      "Iteración: 6\n",
      "Fold accuracy: 0.89\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 1]\n",
      " [3 3 0 0 1 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8913043478260869\n",
      "Precision: 0.9106481481481481\n",
      "Recall: 0.8839285714285715\n",
      "F1-Score: 0.8753812636165578\n",
      "******\n",
      "\n",
      "Iteración: 7\n",
      "Fold accuracy: 0.96\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 0]\n",
      " [1 6 0 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 1 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.9555555555555556\n",
      "Precision: 0.9553571428571429\n",
      "Recall: 0.9553571428571429\n",
      "F1-Score: 0.953968253968254\n",
      "******\n",
      "\n",
      "Iteración: 8\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 0]\n",
      " [3 4 0 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9500000000000001\n",
      "Recall: 0.9285714285714285\n",
      "F1-Score: 0.9251336898395722\n",
      "******\n",
      "\n",
      "Iteración: 9\n",
      "Fold accuracy: 0.80\n",
      "Matriz de Confusión:\n",
      " [[5 2 0 0 0 0]\n",
      " [2 3 0 0 0 3]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [2 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.8\n",
      "Precision: 0.8037037037037037\n",
      "Recall: 0.8065476190476191\n",
      "F1-Score: 0.798736802413273\n",
      "******\n",
      "\n",
      "Iteración: 10\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 0]\n",
      " [3 5 0 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [1 0 0 7 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.9111111111111111\n",
      "Precision: 0.9393939393939394\n",
      "Recall: 0.9166666666666666\n",
      "F1-Score: 0.9133903133903134\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "n_iter = 0 ## Contador número de iteraciones del Cross-Validation\n",
    "# Evaluamos el modelo usando cross-validation\n",
    "for train_index, test_index in skf.split(X_scaled, y_o):\n",
    "    n_iter +=1\n",
    "    print('Iteración:',n_iter)\n",
    "    \n",
    "    ## División del dataset en entrenamiento y test\n",
    "    X_train, X_test = X_scaled[train_index], X_scaled[test_index]\n",
    "    y_train, y_test = y_o[train_index], y_o[test_index]\n",
    "    \n",
    "    ## Entrenar el algoritmo\n",
    "    knn_2.fit(X_train, y_train)\n",
    "    score = knn_2.score(X_test, y_test)\n",
    "    print(\"Fold accuracy: {:.2f}\".format(score))\n",
    "    \n",
    "    ## Predicción del algoritmo\n",
    "    y_pred = knn_2.predict(X_test)\n",
    "    \n",
    "    ## Métricas de evaluación\n",
    "    cm = confusion_matrix(y_test, y_pred)\n",
    "    accu = accuracy_score(y_test, y_pred)\n",
    "    prec = precision_score(y_test, y_pred, average='macro')\n",
    "    recall = recall_score(y_test, y_pred, average='macro')\n",
    "    f1 = f1_score(y_test, y_pred, average='macro')\n",
    "    \n",
    "    ## Guardar resultados métricas de evaluación de cada iteración\n",
    "    cm_list.append(cm)\n",
    "    accu_list.append(accu)\n",
    "    prec_list.append(prec)\n",
    "    recall_list.append(recall)\n",
    "    f1_list.append(f1)\n",
    "    \n",
    "    ## Imprimir Métricas de evaluación en cada iteración\n",
    "    print('Matriz de Confusión:\\n',cm)\n",
    "    print('')\n",
    "    print('Accuracy:',accu)\n",
    "    print('Precision:',prec)\n",
    "    print('Recall:',recall)\n",
    "    print('F1-Score:',f1)\n",
    "    print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed6b4df2-d5bc-492d-9bf2-d295e3cde424",
   "metadata": {},
   "source": [
    "- Podemos observar el resultado de una 10-Fold Stratified Cross-Validation de un clasificador KNN con k=2. En este caso, el conjunto de datos se divide en 10 pliegues o folds y el clasificador KNN se entrena y evalúa en cada pliegue por separado. \n",
    "\n",
    "- Observamos que en todas las iteraciones, el clasificador parece tener una accuracy que oscila entre 0.82 y 0.96, mientras que los valores de precision, recall y F1-score se sitúan en torno a 0.9."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "07fac47b-1ef2-45e0-8834-011630774206",
   "metadata": {},
   "source": [
    "##### Evaluación final del algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "bdfe4ff7-98b2-4800-b60a-12b730fc56f3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 0]\n",
      " [2 4 0 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.893\n",
      "Precision: 0.907\n",
      "Recall: 0.892\n",
      "F1-Score: 0.89\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cm_mean = np.mean(cm_list, axis=0).astype(int)\n",
    "accu_mean = np.mean(accu_list, axis=0)\n",
    "prec_mean = np.mean(prec_list, axis=0)\n",
    "recall_mean = np.mean(recall_list, axis=0)\n",
    "f1_mean = np.mean(f1_list, axis=0)\n",
    "\n",
    "print('Matriz de Confusión:\\n',cm_mean)\n",
    "print('')\n",
    "print('Accuracy:',np.round(accu_mean,3))\n",
    "print('Precision:',np.round(prec_mean,3))\n",
    "print('Recall:',np.round(recall_mean,3))\n",
    "print('F1-Score:',np.round(f1_mean,3))\n",
    "print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d465eb9f-db5a-454a-be89-222d88a6b65d",
   "metadata": {},
   "source": [
    "- Obtenemos el resultado de promediar el rendimiento del clasificador en varias ejecuciones. La matriz de confusión, la accuracy, la precision, el recall y el F1-score se calculan para cada ejecución y, a continuación, se calculan los valores medios.\n",
    "\n",
    "- La matriz de confusión media resultante muestra que la mayoría de las predicciones del clasificador son correctas, ya que la mayoría de las entradas están en la diagonal de la matriz (lo que corresponde a muestras correctamente clasificadas). No obstante hay alguna muestra mal clasificada. \n",
    "\n",
    "- Los valores medios de accuracy, precision, recall y puntuación F1 resultantes son respectivamente 0.893, 0.907, 0.892 y 0.89, lo que implica que, en promedio, el modelo funciona relativamente bien. Sin embargo, algunas clases tienen puntuaciones más bajas que otras, lo que indica que el modelo tiene más dificultades para clasificarlas. \n",
    "\n",
    "- En general, el clasificador tiene un buen rendimiento, pero aún con margen de mejora, especialmente para las clases con valores más bajos de precision, recall y F1-score.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "627cd4d0-02b5-40ad-90fb-b8f17499a6f0",
   "metadata": {},
   "source": [
    "#### Para k=4"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "59b537d8-e8be-4097-89ee-5134ff1d6bf7",
   "metadata": {
    "tags": []
   },
   "source": [
    "##### Preparación algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "b4e4f576-2c8c-4871-9ae3-1e2f63e4580d",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "knn_4 = KNeighborsClassifier(n_neighbors = 4)\n",
    "\n",
    "# 10-Fold Stratified Cross-Validation \n",
    "skf = StratifiedKFold(n_splits=10, random_state=None, shuffle=False)\n",
    "# random_state = None: el generador de números aleatorios utilizará un valor de semilla diferente \n",
    "    # cada vez que se ejecute el código, lo que resultará en un barajado diferente de los datos cada vez.\n",
    "# shuffle=False: los datos se utilizan en el orden del conjunto de datos de entrada y no se barajan.\n",
    "\n",
    "\n",
    "## Arrays para almacenar los resultados de cada iteración\n",
    "cm_list = []\n",
    "accu_list = []\n",
    "prec_list = []\n",
    "recall_list = []\n",
    "f1_list = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d3d513fc-b978-442c-afc6-87979ad6c32c",
   "metadata": {},
   "source": [
    "##### Ejecución algoritmo KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "8526365a-35a4-4220-b04a-593554d3bd98",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteración: 1\n",
      "Fold accuracy: 0.85\n",
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 2]\n",
      " [3 5 0 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 0 6 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [1 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8478260869565217\n",
      "Precision: 0.8872053872053871\n",
      "Recall: 0.8511904761904763\n",
      "F1-Score: 0.8579026752401364\n",
      "******\n",
      "\n",
      "Iteración: 2\n",
      "Fold accuracy: 0.89\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 1]\n",
      " [2 5 1 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [1 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8913043478260869\n",
      "Precision: 0.9106481481481481\n",
      "Recall: 0.8958333333333334\n",
      "F1-Score: 0.8938641695994637\n",
      "******\n",
      "\n",
      "Iteración: 3\n",
      "Fold accuracy: 0.80\n",
      "Matriz de Confusión:\n",
      " [[5 1 0 0 0 2]\n",
      " [2 5 0 0 0 1]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 1 5 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [1 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.8043478260869565\n",
      "Precision: 0.8240740740740741\n",
      "Recall: 0.8035714285714285\n",
      "F1-Score: 0.8045051353874882\n",
      "******\n",
      "\n",
      "Iteración: 4\n",
      "Fold accuracy: 0.89\n",
      "Matriz de Confusión:\n",
      " [[8 0 0 0 0 0]\n",
      " [1 5 0 0 0 2]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [2 0 0 0 0 5]]\n",
      "\n",
      "Accuracy: 0.8913043478260869\n",
      "Precision: 0.906926406926407\n",
      "Recall: 0.8898809523809524\n",
      "F1-Score: 0.8876036244457297\n",
      "******\n",
      "\n",
      "Iteración: 5\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[6 2 0 0 0 0]\n",
      " [1 5 1 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.9130434782608695\n",
      "Precision: 0.9100529100529101\n",
      "Recall: 0.9107142857142857\n",
      "F1-Score: 0.9092436974789916\n",
      "******\n",
      "\n",
      "Iteración: 6\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[8 0 0 0 0 0]\n",
      " [3 3 0 0 1 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.9130434782608695\n",
      "Precision: 0.936026936026936\n",
      "Recall: 0.9047619047619048\n",
      "F1-Score: 0.8972136222910216\n",
      "******\n",
      "\n",
      "Iteración: 7\n",
      "Fold accuracy: 0.89\n",
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 1]\n",
      " [2 4 0 0 0 1]\n",
      " [0 0 6 1 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.8888888888888888\n",
      "Precision: 0.9064814814814816\n",
      "Recall: 0.8809523809523809\n",
      "F1-Score: 0.8800691683044625\n",
      "******\n",
      "\n",
      "Iteración: 8\n",
      "Fold accuracy: 0.82\n",
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 1]\n",
      " [3 1 3 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [1 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8222222222222222\n",
      "Precision: 0.8624999999999999\n",
      "Recall: 0.8125\n",
      "F1-Score: 0.7757352941176471\n",
      "******\n",
      "\n",
      "Iteración: 9\n",
      "Fold accuracy: 0.76\n",
      "Matriz de Confusión:\n",
      " [[3 2 0 0 0 2]\n",
      " [1 3 0 0 1 3]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [2 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.7555555555555555\n",
      "Precision: 0.7534090909090909\n",
      "Recall: 0.7589285714285715\n",
      "F1-Score: 0.7479982006297795\n",
      "******\n",
      "\n",
      "Iteración: 10\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 0]\n",
      " [2 5 0 0 0 1]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444443\n",
      "Recall: 0.9375\n",
      "F1-Score: 0.9309012066365008\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "n_iter = 0 ## Contador número de iteraciones del Cross-Validation\n",
    "\n",
    "# Evaluamos el modelo usando cross-validation\n",
    "for train_index, test_index in skf.split(X_scaled, y_o):\n",
    "    n_iter +=1\n",
    "    print('Iteración:',n_iter)\n",
    "    \n",
    "    ## División del dataset en entrenamiento y test\n",
    "    X_train, X_test = X_scaled[train_index], X_scaled[test_index]\n",
    "    y_train, y_test = y_o[train_index], y_o[test_index]\n",
    "\n",
    "    ## Entrenar el algoritmo\n",
    "    knn_4.fit(X_train, y_train)\n",
    "    score = knn_4.score(X_test, y_test)\n",
    "    print(\"Fold accuracy: {:.2f}\".format(score))\n",
    "    \n",
    "    ## Predicción del algoritmo\n",
    "    y_pred = knn_4.predict(X_test)\n",
    "    \n",
    "    ## Métricas de evaluación\n",
    "    cm = confusion_matrix(y_test, y_pred)\n",
    "    accu = accuracy_score(y_test, y_pred)\n",
    "    prec = precision_score(y_test, y_pred, average='macro')\n",
    "    recall = recall_score(y_test, y_pred, average='macro')\n",
    "    f1 = f1_score(y_test, y_pred, average='macro')\n",
    "    \n",
    "    ## Guardar resultados métricas de evaluación de cada iteración\n",
    "    cm_list.append(cm)\n",
    "    accu_list.append(accu)\n",
    "    prec_list.append(prec)\n",
    "    recall_list.append(recall)\n",
    "    f1_list.append(f1)\n",
    "    \n",
    "    ## Imprimir Métricas de evaluación en cada iteración\n",
    "    print('Matriz de Confusión:\\n',cm)\n",
    "    print('')\n",
    "    print('Accuracy:',accu)\n",
    "    print('Precision:',prec)\n",
    "    print('Recall:',recall)\n",
    "    print('F1-Score:',f1)\n",
    "    print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f78dd0c8-cb8d-4ee0-85db-a97c8247705a",
   "metadata": {},
   "source": [
    "##### Evaluación final del algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "157b7c70-ccfe-4660-826d-a3c42e537a67",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matriz de Confusión:\n",
      " [[6 0 0 0 0 0]\n",
      " [2 4 0 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.866\n",
      "Precision: 0.884\n",
      "Recall: 0.865\n",
      "F1-Score: 0.859\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cm_mean = np.mean(cm_list, axis=0).astype(int)\n",
    "accu_mean = np.mean(accu_list, axis=0)\n",
    "prec_mean = np.mean(prec_list, axis=0)\n",
    "recall_mean = np.mean(recall_list, axis=0)\n",
    "f1_mean = np.mean(f1_list, axis=0)\n",
    "\n",
    "print('Matriz de Confusión:\\n',cm_mean)\n",
    "print('')\n",
    "print('Accuracy:',np.round(accu_mean,3))\n",
    "print('Precision:',np.round(prec_mean,3))\n",
    "print('Recall:',np.round(recall_mean,3))\n",
    "print('F1-Score:',np.round(f1_mean,3))\n",
    "print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e12f21a0-4218-409a-912d-fc6284061245",
   "metadata": {},
   "source": [
    "- Para k=4:\n",
    "    - Misma matriz de confusión que para k=2.\n",
    "    - Observamos que los valores medios de la accuracy, precision, recall y F1-Score son ligeramente peores que para k=2. Por tanto, el rendimiento del modelo no es tan bueno como k=2. Esto puede deberse a que a medida que k aumenta, el modelo tiende a ser más sensible al ruido."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "001303c9-cd3b-4271-bfd9-d569409f6df3",
   "metadata": {},
   "source": [
    "#### Para k=6"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "608f0cc0-1a3d-4ad7-835a-5902dc216895",
   "metadata": {
    "tags": []
   },
   "source": [
    "##### Preparación algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "f2cc10b9-80f1-4039-8a4e-71bc215f8fba",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "knn_6 = KNeighborsClassifier(n_neighbors = 6)\n",
    "\n",
    "# 10-Fold Stratified Cross-Validation \n",
    "skf = StratifiedKFold(n_splits=10, random_state=None, shuffle=False)\n",
    "# random_state = None: el generador de números aleatorios utilizará un valor de semilla diferente \n",
    "    # cada vez que se ejecute el código, lo que resultará en un barajado diferente de los datos cada vez.\n",
    "# shuffle=False: los datos se utilizan en el orden del conjunto de datos de entrada y no se barajan.\n",
    "\n",
    "\n",
    "## Arrays para almacenar los resultados de cada iteración\n",
    "cm_list = []\n",
    "accu_list = []\n",
    "prec_list = []\n",
    "recall_list = []\n",
    "f1_list = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0c699ed-ebce-419b-b7ae-7b252d8037b7",
   "metadata": {},
   "source": [
    "##### Ejecución algoritmo KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "496e3c3e-ac3a-47e6-9093-abb7b9c298bb",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteración: 1\n",
      "Fold accuracy: 0.83\n",
      "Matriz de Confusión:\n",
      " [[5 1 0 0 0 2]\n",
      " [3 5 0 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 0 6 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [1 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8260869565217391\n",
      "Precision: 0.851851851851852\n",
      "Recall: 0.8303571428571429\n",
      "F1-Score: 0.836074600780483\n",
      "******\n",
      "\n",
      "Iteración: 2\n",
      "Fold accuracy: 0.83\n",
      "Matriz de Confusión:\n",
      " [[5 2 0 0 0 1]\n",
      " [2 5 1 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 0 6 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [1 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8260869565217391\n",
      "Precision: 0.8389550264550264\n",
      "Recall: 0.8303571428571429\n",
      "F1-Score: 0.8323592257415786\n",
      "******\n",
      "\n",
      "Iteración: 3\n",
      "Fold accuracy: 0.83\n",
      "Matriz de Confusión:\n",
      " [[5 1 0 0 0 2]\n",
      " [3 5 0 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [1 0 1 5 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8260869565217391\n",
      "Precision: 0.8425925925925926\n",
      "Recall: 0.8273809523809524\n",
      "F1-Score: 0.8253384687208216\n",
      "******\n",
      "\n",
      "Iteración: 4\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[8 0 0 0 0 0]\n",
      " [1 5 0 0 0 2]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [1 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.9130434782608695\n",
      "Precision: 0.9249999999999999\n",
      "Recall: 0.9136904761904762\n",
      "F1-Score: 0.9096866096866097\n",
      "******\n",
      "\n",
      "Iteración: 5\n",
      "Fold accuracy: 0.83\n",
      "Matriz de Confusión:\n",
      " [[4 3 0 0 0 1]\n",
      " [1 4 2 0 0 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 1 0 7 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.8260869565217391\n",
      "Precision: 0.8291666666666666\n",
      "Recall: 0.8244047619047619\n",
      "F1-Score: 0.8173789173789174\n",
      "******\n",
      "\n",
      "Iteración: 6\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[8 0 0 0 0 0]\n",
      " [3 3 0 0 1 0]\n",
      " [0 0 8 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 7]]\n",
      "\n",
      "Accuracy: 0.9130434782608695\n",
      "Precision: 0.936026936026936\n",
      "Recall: 0.9047619047619048\n",
      "F1-Score: 0.8972136222910216\n",
      "******\n",
      "\n",
      "Iteración: 7\n",
      "Fold accuracy: 0.87\n",
      "Matriz de Confusión:\n",
      " [[5 0 0 0 0 2]\n",
      " [2 4 0 0 0 1]\n",
      " [0 0 6 1 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.8666666666666667\n",
      "Precision: 0.8884078884078884\n",
      "Recall: 0.8571428571428571\n",
      "F1-Score: 0.8579861830635824\n",
      "******\n",
      "\n",
      "Iteración: 8\n",
      "Fold accuracy: 0.82\n",
      "Matriz de Confusión:\n",
      " [[7 0 0 0 0 0]\n",
      " [3 1 3 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 8 0]\n",
      " [2 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.8222222222222222\n",
      "Precision: 0.8805555555555555\n",
      "Recall: 0.8154761904761904\n",
      "F1-Score: 0.7779190623617868\n",
      "******\n",
      "\n",
      "Iteración: 9\n",
      "Fold accuracy: 0.73\n",
      "Matriz de Confusión:\n",
      " [[2 3 0 0 0 2]\n",
      " [1 3 0 0 1 3]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [2 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.7333333333333333\n",
      "Precision: 0.7200757575757576\n",
      "Recall: 0.7351190476190476\n",
      "F1-Score: 0.7211361737677527\n",
      "******\n",
      "\n",
      "Iteración: 10\n",
      "Fold accuracy: 0.91\n",
      "Matriz de Confusión:\n",
      " [[6 1 0 0 0 0]\n",
      " [1 5 0 0 0 2]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 8 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 8]]\n",
      "\n",
      "Accuracy: 0.9111111111111111\n",
      "Precision: 0.915079365079365\n",
      "Recall: 0.9136904761904763\n",
      "F1-Score: 0.9100529100529101\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "n_iter = 0 ## Contador número de iteraciones del Cross-Validation\n",
    "\n",
    "# Evaluamos el modelo usando cross-validation\n",
    "for train_index, test_index in skf.split(X_scaled, y_o):\n",
    "    \n",
    "    n_iter +=1\n",
    "    print('Iteración:',n_iter)\n",
    "    \n",
    "    ## División del dataset en entrenamiento y test\n",
    "    X_train, X_test = X_scaled[train_index], X_scaled[test_index]\n",
    "    y_train, y_test = y_o[train_index], y_o[test_index]\n",
    "    \n",
    "    ## Entrenar el algoritmo\n",
    "    knn_6.fit(X_train, y_train)\n",
    "    score = knn_6.score(X_test, y_test)\n",
    "    print(\"Fold accuracy: {:.2f}\".format(score))\n",
    "    \n",
    "    ## Predicción del algoritmo\n",
    "    y_pred = knn_6.predict(X_test)\n",
    "    \n",
    "    ## Métricas de evaluación\n",
    "    cm = confusion_matrix(y_test, y_pred)\n",
    "    accu = accuracy_score(y_test, y_pred)\n",
    "    prec = precision_score(y_test, y_pred, average='macro')\n",
    "    recall = recall_score(y_test, y_pred, average='macro')\n",
    "    f1 = f1_score(y_test, y_pred, average='macro')\n",
    "    \n",
    "    ## Guardar resultados métricas de evaluación de cada iteración\n",
    "    cm_list.append(cm)\n",
    "    accu_list.append(accu)\n",
    "    prec_list.append(prec)\n",
    "    recall_list.append(recall)\n",
    "    f1_list.append(f1)\n",
    "    \n",
    "    ## Imprimir Métricas de evaluación en cada iteración\n",
    "    print('Matriz de Confusión:\\n',cm)\n",
    "    print('')\n",
    "    print('Accuracy:',accu)\n",
    "    print('Precision:',prec)\n",
    "    print('Recall:',recall)\n",
    "    print('F1-Score:',f1)\n",
    "    print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6427c7e-8517-4045-9779-0302d5ef5f38",
   "metadata": {},
   "source": [
    "##### Evaluación final del algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "938c048a-ad2c-4d9f-9941-dc7873b0ea6a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matriz de Confusión:\n",
      " [[5 1 0 0 0 1]\n",
      " [2 4 0 0 0 0]\n",
      " [0 0 7 0 0 0]\n",
      " [0 0 0 7 0 0]\n",
      " [0 0 0 0 7 0]\n",
      " [0 0 0 0 0 6]]\n",
      "\n",
      "Accuracy: 0.846\n",
      "Precision: 0.863\n",
      "Recall: 0.845\n",
      "F1-Score: 0.839\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cm_mean = np.mean(cm_list, axis=0).astype(int)\n",
    "accu_mean = np.mean(accu_list, axis=0)\n",
    "prec_mean = np.mean(prec_list, axis=0)\n",
    "recall_mean = np.mean(recall_list, axis=0)\n",
    "f1_mean = np.mean(f1_list, axis=0)\n",
    "\n",
    "print('Matriz de Confusión:\\n',cm_mean)\n",
    "print('')\n",
    "print('Accuracy:',np.round(accu_mean,3))\n",
    "print('Precision:',np.round(prec_mean,3))\n",
    "print('Recall:',np.round(recall_mean,3))\n",
    "print('F1-Score:',np.round(f1_mean,3))\n",
    "print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2d06bb0-312c-4b6f-bda9-cb3017b6b898",
   "metadata": {},
   "source": [
    "- Para k=6, observamos que los valores medios de la accuracy, precision, recall y F1-Score son ligeramente peores que para k=2 y k=4. Por tanto, el rendimiento del modelo ha disminuido respecto a k=2 y k=4. Esto puede deberse a que a medida que k aumenta, el modelo tiende a ser más sensible al ruido.\n",
    "\n",
    "- k=6 no es un valor óptimo. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "40e992a6-8d10-4588-bc4d-bcbe5e3a045c",
   "metadata": {},
   "source": [
    "**Conclusión**\n",
    "- En conclusión, podemos ver que el rendimiento del clasificador disminuye a medida que aumenta el valor de k.  \n",
    "- En general, el clasificador obtiene mejores resultados con un valor menor de k. \n",
    "    - Una posible razón de ello podría ser que el uso de un valor menor de k permite al clasificador identificar mejor las instancias más cercanas y similares a una instancia desconocida, lo que a su vez aumenta las posibilidades de clasificarla correctamente. \n",
    "    - Podría ser que a medida que k aumenta, el modelo tiende a infraajustarse (underfitting), es decir, es menos capaz de ajustarse a los datos, lo que se traduce en una disminución del rendimiento.\n",
    "    - También es posible que el sobremuestreo con SMOTE haya influido en el rendimiento del clasificador KNN, provocando más ruido en el conjunto de datos, lo que podría explicar el mejor rendimiento de un valor k más bajo. Además, el sobremuestreo de muestras sintéticas podría ser menos informativo que las muestras reales, lo que también podría disminuir el rendimiento.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d49957f1-5597-4128-a5f3-4c9a21c04ea5",
   "metadata": {},
   "source": [
    "## iris.arff"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "390cb38d-1dde-42d0-844c-6964e121d771",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# cargamos el .arff file\n",
    "iris = arff.loadarff('iris.arff')\n",
    "# iris es una tupla que contiene dos elementos: un array numpy y una lista de tuplas\n",
    "# primer elemento de la tupla (el array numpy) es el que nos interesa (iris[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "743466ea-8011-4f04-b72a-7ea721018a20",
   "metadata": {},
   "outputs": [],
   "source": [
    "# convertimos los datos a df\n",
    "iris_df = pd.DataFrame(iris[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "0394f60c-42f1-4a4e-8374-24d9780a480b",
   "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>sepallength</th>\n",
       "      <th>sepalwidth</th>\n",
       "      <th>petallength</th>\n",
       "      <th>petalwidth</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>b'Iris-setosa'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>b'Iris-setosa'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>b'Iris-setosa'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.6</td>\n",
       "      <td>3.1</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.2</td>\n",
       "      <td>b'Iris-setosa'</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>b'Iris-setosa'</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   sepallength  sepalwidth  petallength  petalwidth           class\n",
       "0          5.1         3.5          1.4         0.2  b'Iris-setosa'\n",
       "1          4.9         3.0          1.4         0.2  b'Iris-setosa'\n",
       "2          4.7         3.2          1.3         0.2  b'Iris-setosa'\n",
       "3          4.6         3.1          1.5         0.2  b'Iris-setosa'\n",
       "4          5.0         3.6          1.4         0.2  b'Iris-setosa'"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 5 primeras filas del df\n",
    "iris_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "4a26a56f-76ad-44fe-a274-ca987c8415d7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([b'Iris-setosa', b'Iris-versicolor', b'Iris-virginica'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# nombres de las clases\n",
    "iris_df['class'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "82d4b1b6-dd68-47c0-bd0c-2177cd0b41cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['iris_setosa', 'iris_versicolor', 'iris_virginica'], dtype=object)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# renombramos las clases para un mejor entendimiento \n",
    "iris_df['class'].replace({b'Iris-setosa': 'iris_setosa',\n",
    "                          b'Iris-versicolor': 'iris_versicolor',\n",
    "                          b'Iris-virginica': 'iris_virginica'\n",
    "                        }, inplace = True)\n",
    "iris_df['class'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d37497dc-060d-4ec3-9306-e80233f26e20",
   "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>sepallength</th>\n",
       "      <th>sepalwidth</th>\n",
       "      <th>petallength</th>\n",
       "      <th>petalwidth</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.1</td>\n",
       "      <td>3.5</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>iris_setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4.9</td>\n",
       "      <td>3.0</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>iris_setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.7</td>\n",
       "      <td>3.2</td>\n",
       "      <td>1.3</td>\n",
       "      <td>0.2</td>\n",
       "      <td>iris_setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.6</td>\n",
       "      <td>3.1</td>\n",
       "      <td>1.5</td>\n",
       "      <td>0.2</td>\n",
       "      <td>iris_setosa</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>1.4</td>\n",
       "      <td>0.2</td>\n",
       "      <td>iris_setosa</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   sepallength  sepalwidth  petallength  petalwidth        class\n",
       "0          5.1         3.5          1.4         0.2  iris_setosa\n",
       "1          4.9         3.0          1.4         0.2  iris_setosa\n",
       "2          4.7         3.2          1.3         0.2  iris_setosa\n",
       "3          4.6         3.1          1.5         0.2  iris_setosa\n",
       "4          5.0         3.6          1.4         0.2  iris_setosa"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 5 primeras filas del df correcto\n",
    "iris_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "dfec26d3-c4d4-4def-8ebd-1aa05a6b7edd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 150 entries, 0 to 149\n",
      "Data columns (total 5 columns):\n",
      " #   Column       Non-Null Count  Dtype  \n",
      "---  ------       --------------  -----  \n",
      " 0   sepallength  150 non-null    float64\n",
      " 1   sepalwidth   150 non-null    float64\n",
      " 2   petallength  150 non-null    float64\n",
      " 3   petalwidth   150 non-null    float64\n",
      " 4   class        150 non-null    object \n",
      "dtypes: float64(4), object(1)\n",
      "memory usage: 6.0+ KB\n"
     ]
    }
   ],
   "source": [
    "# info de las columnas del df\n",
    "iris_df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "533535d5-e973-464a-87dd-98e6437c09d0",
   "metadata": {},
   "source": [
    "- Observamos que el iris_df tiene 150 filas y 5 columnas. \n",
    "- Los tipos de datos de las columnas son float64 para 4 de las columnas y object para la columna class. \n",
    "- Los nombres de las columnas son \"sepallength\", \"sepalwidth\", \"petallength\", \"petalwidth\", y \"class\".\n",
    "- Todas las columnas tienen 150 valores no nulos, lo que significa que no hay valores perdidos (missing values) en el df."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "16cc73f5-a064-4367-a3a2-4dcdbeb5ea47",
   "metadata": {},
   "source": [
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "96e777f2-c923-467e-b2d4-a2585f5061fd",
   "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>sepallength</th>\n",
       "      <th>sepalwidth</th>\n",
       "      <th>petallength</th>\n",
       "      <th>petalwidth</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "      <td>150.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>5.843333</td>\n",
       "      <td>3.054000</td>\n",
       "      <td>3.758667</td>\n",
       "      <td>1.198667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.828066</td>\n",
       "      <td>0.433594</td>\n",
       "      <td>1.764420</td>\n",
       "      <td>0.763161</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>4.300000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>5.100000</td>\n",
       "      <td>2.800000</td>\n",
       "      <td>1.600000</td>\n",
       "      <td>0.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>5.800000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>4.350000</td>\n",
       "      <td>1.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>6.400000</td>\n",
       "      <td>3.300000</td>\n",
       "      <td>5.100000</td>\n",
       "      <td>1.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>7.900000</td>\n",
       "      <td>4.400000</td>\n",
       "      <td>6.900000</td>\n",
       "      <td>2.500000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       sepallength  sepalwidth  petallength  petalwidth\n",
       "count   150.000000  150.000000   150.000000  150.000000\n",
       "mean      5.843333    3.054000     3.758667    1.198667\n",
       "std       0.828066    0.433594     1.764420    0.763161\n",
       "min       4.300000    2.000000     1.000000    0.100000\n",
       "25%       5.100000    2.800000     1.600000    0.300000\n",
       "50%       5.800000    3.000000     4.350000    1.300000\n",
       "75%       6.400000    3.300000     5.100000    1.800000\n",
       "max       7.900000    4.400000     6.900000    2.500000"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# estadísticas descriptivas \n",
    "iris_df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c30d202f-6bf5-4cc6-9d28-02a6853ac540",
   "metadata": {},
   "source": [
    "- El método df.describe() genera estadísticas descriptivas de un df. Obtenemos el recuento, la media, la desviación estándar, el mínimo, el primer cuartil (25%), la mediana (50%), el tercer cuartil (75%) y el máximo de cada columna.\n",
    "\n",
    "- Esta información puede ser útil para comprender la distribución y el rango de valores de cada columna del df.\n",
    "\n",
    "- El iris datsaset contiene mediciones de 4 características (sepallength, sepalwidth, petallength y petalwidth) de 150 flores de Iris diferentes. Estas características se utilizan para distinguir entre 3 especies diferentes de Iris: Iris setosa, Iris virginica e Iris versicolor.\n",
    "- Observando las estadísticas resumidas del conjunto de datos, podemos observar, por ejemplo, que los valores medios de las cuatro características están próximos a la mediana (50%), lo que indica que los datos son bastante simétricos. \n",
    "\n",
    "- En general, el iris_df parece bastante limpio y con un buen comportamiento, sin problemas importantes como falta de datos o valores atípicos. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d00e93d5-eb0a-4223-873c-f3344d56073b",
   "metadata": {},
   "source": [
    "### Distribución de cada variable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "c0fe3190-fcf1-4db7-a22f-8cd8ee3eeb44",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2500x1500 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# creating a figure and a 3x3 grid of subplots\n",
    "fig, axs = plt.subplots(nrows=2, ncols=2, figsize=(25, 15))\n",
    "\n",
    "# seaborn's histplot to plot histograms on the subplots\n",
    "sns.histplot(data=iris_df, x='sepallength', element='step', stat='density', ax=axs[0, 0])\n",
    "sns.histplot(data=iris_df, x='sepalwidth', element='step', stat='density', ax=axs[0, 1])\n",
    "sns.histplot(data=iris_df, x='petallength', element='step', stat='density', ax=axs[1, 1])\n",
    "sns.histplot(data=iris_df, x='petalwidth', element='step', stat='density', ax=axs[1, 0])\n",
    "\n",
    "# axis labels\n",
    "axs[0, 0].set(xlabel='sepallength', ylabel='Density')\n",
    "axs[0, 1].set(xlabel='sepalwidth', ylabel='Density')\n",
    "axs[1, 1].set(xlabel='petallength', ylabel='Density')\n",
    "axs[1, 0].set(xlabel='petalwidth', ylabel='Density')\n",
    "\n",
    "# adjusting the spacing between the subplots\n",
    "plt.tight_layout()\n",
    "\n",
    "# showing the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fcb96795-641d-4f6d-804a-4b9c2d6d248b",
   "metadata": {},
   "source": [
    "- Observamos que sepallegnth y sepalwidth se distribuyen de manera más normal que petalwidth y petallength, que tienen una asimetría positiva. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe4271cb-71d1-465c-bcd8-0a89727581f7",
   "metadata": {},
   "source": [
    "### Comprobación del desequilibrio de clases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "49e6e3ab-2502-4217-8804-d16a28a15f13",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iris_setosa        50\n",
      "iris_versicolor    50\n",
      "iris_virginica     50\n",
      "Name: class, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# Checking data imbalance \n",
    "print(iris_df['class'].value_counts()) # recuento de instancias de cada clase en el df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "20aa9fd8-66da-4498-962a-0dacba91b84c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot del count\n",
    "sns.countplot(x = iris_df['class'], color = 'blue')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "216a8884-33bc-4489-8bd5-a693b0320196",
   "metadata": {},
   "source": [
    "- Podemos observar que el iris_df está bien equilibrado, con el mismo número de instancias (50) para cada una de las tres especies de Iris: Iris setosa, Iris versicolor e Iris virginica. \n",
    "- Esto significa que los datos se distribuyen uniformemente entre las tres clases y que no hay desequilibrio de clases (dataset balanceado).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d09dfd6c-8665-41e8-a39e-0e9f2d3be435",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Separar las variables independientes de la variable dependiente\n",
    "- Separar en atributos (X) y etiquetas (y).\n",
    "- Las columnas del conjunto de datos se pueden separar en feature columns (X) y target column (y)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "c3ce9e63-ce28-480a-82c6-e39b0884b134",
   "metadata": {},
   "outputs": [],
   "source": [
    "# variables independientes (feature columns): las variables que se utilizan para predecir la variable objetivo (y)\n",
    "X_iris = iris_df.iloc[:, :-1].values\n",
    "# variable dependiente (target column): en nuestro caso 'class'\n",
    "y_iris = iris_df.iloc[:, -1].values"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06304730-479c-4e85-91d1-baa6f59a19d1",
   "metadata": {},
   "source": [
    "### Escalar los atributos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "5373e97d-2169-49d1-8310-1304c9a20196",
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler()\n",
    "scaler.fit(X_iris)\n",
    "\n",
    "X_scaled_iris = scaler.transform(X_iris)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e6d1ef48-b835-49d1-85f9-5923b33dfd0e",
   "metadata": {},
   "source": [
    "- Escalamos para que el algoritmo le dé la misma importancia a todos los datos.\n",
    "- Algoritmos como KNN se basan en la distancia entre los puntos de datos, por lo que es importante que las features estén en la misma escala.\n",
    "- El resultado es un nuevo conjunto de valores que tienen una distribución normal estándar (una distribución gaussiana con una media de 0 y una desviación estándar de 1).\n",
    "\n",
    "- Ahora los tenemos todos dentro de un mismo rango o intervalo.\n",
    "- La variable X_scaled contiene la versión estandarizada de la feature data almacenada en X.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6bb8e228-0ce5-47ae-930b-d77b5f09e32e",
   "metadata": {},
   "source": [
    "### Algoritmo KNN"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "230fd8ff-7ef8-43e8-bf8b-050f625ed027",
   "metadata": {},
   "source": [
    "#### Para k=2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "211cfe01-de62-4645-88fc-e1dd19091040",
   "metadata": {
    "tags": []
   },
   "source": [
    "##### Preparación algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "58caf194-6ac1-49bb-8fc3-52dacfc0da64",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "knn_2_i = KNeighborsClassifier(n_neighbors = 2)\n",
    "\n",
    "# 10-Fold Stratified Cross-Validation \n",
    "skf_i = StratifiedKFold(n_splits=10, random_state=None, shuffle=False)\n",
    "# random_state = None: el generador de números aleatorios utilizará un valor de semilla diferente \n",
    "    # cada vez que se ejecute el código, lo que resultará en un barajado diferente de los datos cada vez.\n",
    "# shuffle=False: los datos se utilizan en el orden del conjunto de datos de entrada y no se barajan.\n",
    "\n",
    "\n",
    "## Arrays para almacenar los resultados de cada iteración\n",
    "cm_list_i = []\n",
    "accu_list_i = []\n",
    "prec_list_i = []\n",
    "recall_list_i = []\n",
    "f1_list_i = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9a32e483-a77d-4120-ab75-906c03960e2f",
   "metadata": {},
   "source": [
    "- Cuando random_state se establece en None y el parámetro shuffle es False, los datos siempre se dividen tal cual, sin barajar ni aleatorizar los datos. Esto significa que las observaciones se utilizan en el orden del conjunto de datos de entrada.\n",
    "- Por tanto, en este caso, cuando random_state=None y shuffle=False, significa que no hay aleatorización al crear los pliegues (folds), los datos no se barajan y las observaciones se utilizan en el orden del conjunto de datos de entrada. El resultado será el mismo cada vez que se ejecute el código.\n",
    "- Esto puede ser útil queremos comparar los resultados del modelo bajo diferentes condiciones, o si queremos asegurar la reproducibilidad de los resultados (queremos que utilice la misma división cada vez que se ejecute el código).\n",
    "- No obstante, dejando de lado el ejercicio, es bueno que vaya cambiando, ya que si vemos que el cojunto de training y test varían mucho, no tenemos datos estables para el problema que queremos resolver."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31599f25-ef05-4521-9a0f-1896d4bbd3af",
   "metadata": {},
   "source": [
    "##### Ejecución algoritmo KNN"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7c3cb337-e4c7-4441-91ba-a3fd763feec9",
   "metadata": {},
   "source": [
    "- Se trata de un problema de clasificación multiclase.\n",
    "- average='macro' calculará la métrica independientemente para cada clase y luego sacará la media. Las métricas macro promediadas darán el mismo peso a cada clase, independientemente del desequilibrio de clases.\n",
    "- Dado que el conjunto de datos Iris está equilibrado, tanto micro como macro darán resultados similares. En este caso usaremos la métrica macro-promediada. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "ed313e33-1665-4b3b-897b-a8e28051fafc",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteración: 1\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 2\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 3\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 4\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 5\n",
      "Fold accuracy: 0.87\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 3 2]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 0.8666666666666667\n",
      "Precision: 0.9047619047619048\n",
      "Recall: 0.8666666666666667\n",
      "F1-Score: 0.861111111111111\n",
      "******\n",
      "\n",
      "Iteración: 6\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 7\n",
      "Fold accuracy: 0.80\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 1]\n",
      " [0 2 3]]\n",
      "\n",
      "Accuracy: 0.8\n",
      "Precision: 0.8055555555555555\n",
      "Recall: 0.7999999999999999\n",
      "F1-Score: 0.7979797979797979\n",
      "******\n",
      "\n",
      "Iteración: 8\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 9\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 10\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "n_iter_i = 0 ## Contador número de iteraciones del Cross-Validation\n",
    "\n",
    "# Evaluamos el modelo usando cross-validation\n",
    "for train_index_i, test_index_i in skf_i.split(X_scaled_iris, y_iris):\n",
    "    \n",
    "    n_iter_i +=1\n",
    "    print('Iteración:',n_iter_i)\n",
    "    \n",
    "    ## División del dataset en entrenamiento y test\n",
    "    X_train_i, X_test_i = X_scaled_iris[train_index_i], X_scaled_iris[test_index_i]\n",
    "    y_train_i, y_test_i = y_iris[train_index_i], y_iris[test_index_i]\n",
    "    \n",
    "    ## Entrenar el algoritmo\n",
    "    knn_2_i.fit(X_train_i, y_train_i)\n",
    "    score_i = knn_2_i.score(X_test_i, y_test_i)\n",
    "    print(\"Fold accuracy: {:.2f}\".format(score_i))\n",
    "    \n",
    "    ## Predicción del algoritmo\n",
    "    y_pred_i = knn_2_i.predict(X_test_i)\n",
    "    \n",
    "    ## Métricas de evaluación\n",
    "    cm_i = confusion_matrix(y_test_i, y_pred_i)\n",
    "    accu_i = accuracy_score(y_test_i, y_pred_i)\n",
    "    prec_i = precision_score(y_test_i, y_pred_i, average='macro')\n",
    "    recall_i = recall_score(y_test_i, y_pred_i, average='macro')\n",
    "    f1_i = f1_score(y_test_i, y_pred_i, average='macro')\n",
    "    \n",
    "    ## Guardar resultados métricas de evaluación de cada iteración\n",
    "    cm_list_i.append(cm_i)\n",
    "    accu_list_i.append(accu_i)\n",
    "    prec_list_i.append(prec_i)\n",
    "    recall_list_i.append(recall_i)\n",
    "    f1_list_i.append(f1_i)\n",
    "    \n",
    "    ## Imprimir Métricas de evaluación en cada iteración\n",
    "    print('Matriz de Confusión:\\n',cm_i)\n",
    "    print('')\n",
    "    print('Accuracy:',accu_i)\n",
    "    print('Precision:',prec_i)\n",
    "    print('Recall:',recall_i)\n",
    "    print('F1-Score:',f1_i)\n",
    "    print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "77efb8b9-6335-46d8-9006-97d59a5284c8",
   "metadata": {},
   "source": [
    "- Podemos observar el resultado de una 10-Fold Stratified Cross-Validation de un clasificador KNN con k=2. En este caso, el conjunto de datos se divide en 10 pliegues o folds y el clasificador KNN se entrena y evalúa en cada pliegue por separado. \n",
    "\n",
    "- La accuracy de todas las iteraciones se sitúa entre 0.87 y 1, lo que sugiere que el clasificador KNN es capaz de predecir correctamente la clase de la mayoría de las muestras del dataset.\n",
    "\n",
    "- La precision, el recall y la F1-score también son elevadas, y en el peor de los casos se sitúan en torno a 0.8. Estas tres métricas indican que el clasificador es capaz de clasificar correctamente un alto porcentaje de los casos positivos y que no clasifica erróneamente como positivos muchos de los casos negativos.\n",
    "\n",
    "- Las matrices de confusión muestran que el clasificador funcionó bien en iris setosa. Sin embargo, en algunas iteraciones, el clasificador clasifica erróneamente iris versicolor e iris virginica. Una posible razón podría ser que estas dos clases son más difíciles de separar. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fc7cdc07-0dd1-4130-b47b-96be286ba243",
   "metadata": {},
   "source": [
    "##### Evaluación final del algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "bfc2adee-3eb1-4331-bca8-17404bfb80d0",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 0]\n",
      " [0 0 4]]\n",
      "\n",
      "Accuracy: 0.953\n",
      "Precision: 0.96\n",
      "Recall: 0.953\n",
      "F1-Score: 0.952\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cm_mean_i = np.mean(cm_list_i, axis=0).astype(int)\n",
    "accu_mean_i = np.mean(accu_list_i, axis=0)\n",
    "prec_mean_i = np.mean(prec_list_i, axis=0)\n",
    "recall_mean_i = np.mean(recall_list_i, axis=0)\n",
    "f1_mean_i = np.mean(f1_list_i, axis=0)\n",
    "\n",
    "print('Matriz de Confusión:\\n',cm_mean_i)\n",
    "print('')\n",
    "print('Accuracy:',np.round(accu_mean_i,3))\n",
    "print('Precision:',np.round(prec_mean_i,3))\n",
    "print('Recall:',np.round(recall_mean_i,3))\n",
    "print('F1-Score:',np.round(f1_mean_i,3))\n",
    "print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c64473c9-7847-4557-9886-3976641330d3",
   "metadata": {},
   "source": [
    "- Obtenemos el resultado de promediar el rendimiento del clasificador en varias ejecuciones. La matriz de confusión, la accuracy, la precision, el recall y el F1-score se calculan para cada ejecución y, a continuación, se calculan los valores medios.\n",
    "\n",
    "- La matriz de confusión muestra que el clasificador funciona bien en la mayoría de los casos. \n",
    "\n",
    "- Las métricas generales de rendimiento son superiores a 0.95, lo que sugiere que el clasificador es capaz de predecir la clase correcta para un gran porcentaje de las muestras.\n",
    "\n",
    "- El modelo tiene una accuracy del 0.953, lo que sugiere que el modelo clasificó correctamente el 95.3% de las muestras. \n",
    "\n",
    "- La precisión del modelo es de 0.96, lo que significa que de todas las muestras clasificadas como positivas, el 96% son verdaderos positivos.\n",
    "\n",
    "- El recall del modelo es de 0.953, lo significa que de todas las muestras del iris_df, el modelo fue capaz de identificar correctamente el 95.3% de ellas.\n",
    "\n",
    "- El F1-score de 0.952 indica un equilibrio entre la precision y el recall del modelo."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "716f3a59-f97f-4198-ad49-7195b2de1a6f",
   "metadata": {},
   "source": [
    "#### Para k=4"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "10690b94-272e-40a3-b1ce-dcc73546686d",
   "metadata": {
    "tags": []
   },
   "source": [
    "##### Preparación algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "9228bb2e-466c-452c-9eb1-1c12ca7b2c03",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "knn_4_i = KNeighborsClassifier(n_neighbors = 4)\n",
    "\n",
    "# 10-Fold Stratified Cross-Validation \n",
    "skf_i = StratifiedKFold(n_splits=10, random_state=None, shuffle=False)\n",
    "# random_state = None: el generador de números aleatorios utilizará un valor de semilla diferente \n",
    "    # cada vez que se ejecute el código, lo que resultará en un barajado diferente de los datos cada vez.\n",
    "# shuffle=False: los datos se utilizan en el orden del conjunto de datos de entrada y no se barajan.\n",
    "\n",
    "\n",
    "## Arrays para almacenar los resultados de cada iteración\n",
    "cm_list_i = []\n",
    "accu_list_i = []\n",
    "prec_list_i = []\n",
    "recall_list_i = []\n",
    "f1_list_i = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0e72ec1b-14a1-4245-99ab-f52eb269779d",
   "metadata": {},
   "source": [
    "##### Ejecución algoritmo KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "9e85cfa1-3661-449e-ae6a-bce94486c406",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteración: 1\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 2\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 3\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 4\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 5\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 1]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 6\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 7\n",
      "Fold accuracy: 0.80\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 1]\n",
      " [0 2 3]]\n",
      "\n",
      "Accuracy: 0.8\n",
      "Precision: 0.8055555555555555\n",
      "Recall: 0.7999999999999999\n",
      "F1-Score: 0.7979797979797979\n",
      "******\n",
      "\n",
      "Iteración: 8\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 9\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 10\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "n_iter_i = 0 ## Contador número de iteraciones del Cross-Validation\n",
    "\n",
    "# Evaluamos el modelo usando cross-validation\n",
    "for train_index_i, test_index_i in skf_i.split(X_scaled_iris, y_iris):\n",
    "    \n",
    "    n_iter_i +=1\n",
    "    print('Iteración:',n_iter_i)\n",
    "    \n",
    "    ## División del dataset en entrenamiento y test\n",
    "    X_train_i, X_test_i = X_scaled_iris[train_index_i], X_scaled_iris[test_index_i]\n",
    "    y_train_i, y_test_i = y_iris[train_index_i], y_iris[test_index_i]\n",
    "\n",
    "    ## Entrenar el algoritmo\n",
    "    knn_4_i.fit(X_train_i, y_train_i)\n",
    "    score_i = knn_4_i.score(X_test_i, y_test_i)\n",
    "    print(\"Fold accuracy: {:.2f}\".format(score_i))\n",
    "    \n",
    "    ## Predicción del algoritmo\n",
    "    y_pred_i = knn_4_i.predict(X_test_i)\n",
    "    \n",
    "    ## Métricas de evaluación\n",
    "    cm_i = confusion_matrix(y_test_i, y_pred_i)\n",
    "    accu_i = accuracy_score(y_test_i, y_pred_i)\n",
    "    prec_i = precision_score(y_test_i, y_pred_i, average='macro')\n",
    "    recall_i = recall_score(y_test_i, y_pred_i, average='macro')\n",
    "    f1_i = f1_score(y_test_i, y_pred_i, average='macro')\n",
    "    \n",
    "    ## Guardar resultados métricas de evaluación de cada iteración\n",
    "    cm_list_i.append(cm_i)\n",
    "    accu_list_i.append(accu_i)\n",
    "    prec_list_i.append(prec_i)\n",
    "    recall_list_i.append(recall_i)\n",
    "    f1_list_i.append(f1_i)\n",
    "    \n",
    "    ## Imprimir Métricas de evaluación en cada iteración\n",
    "    print('Matriz de Confusión:\\n',cm_i)\n",
    "    print('')\n",
    "    print('Accuracy:',accu_i)\n",
    "    print('Precision:',prec_i)\n",
    "    print('Recall:',recall_i)\n",
    "    print('F1-Score:',f1_i)\n",
    "    print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d1f0ab6e-bd4f-4c1e-accd-492f6c9c3ceb",
   "metadata": {},
   "source": [
    "##### Evaluación final del algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "c558f6ab-c2db-4d45-824c-0486b09105de",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 0]\n",
      " [0 0 4]]\n",
      "\n",
      "Accuracy: 0.947\n",
      "Precision: 0.953\n",
      "Recall: 0.947\n",
      "F1-Score: 0.946\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cm_mean_i = np.mean(cm_list_i, axis=0).astype(int)\n",
    "accu_mean_i = np.mean(accu_list_i, axis=0)\n",
    "prec_mean_i = np.mean(prec_list_i, axis=0)\n",
    "recall_mean_i = np.mean(recall_list_i, axis=0)\n",
    "f1_mean_i = np.mean(f1_list_i, axis=0)\n",
    "\n",
    "print('Matriz de Confusión:\\n',cm_mean_i)\n",
    "print('')\n",
    "print('Accuracy:',np.round(accu_mean_i,3))\n",
    "print('Precision:',np.round(prec_mean_i,3))\n",
    "print('Recall:',np.round(recall_mean_i,3))\n",
    "print('F1-Score:',np.round(f1_mean_i,3))\n",
    "print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a886eca7-528c-4ada-b477-909e20f42d64",
   "metadata": {},
   "source": [
    "- Para k=4:\n",
    "    - Misma matriz de confusión que para k=2. \n",
    "    - Observamos que los valores medios de la accuracy, precision, recall y F1-Score son ligeramente peores que para k=2. Por tanto, el rendimiento del modelo no es tan bueno como k=2. Esto puede deberse a que a medida que k aumenta, el modelo tiende a ser más sensible al ruido.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a4a98f8d-bac8-431a-a552-f6fd918f8cbc",
   "metadata": {},
   "source": [
    "#### Para k=6"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c2d3cbdd-d557-4972-93c7-808762d2696e",
   "metadata": {
    "tags": []
   },
   "source": [
    "##### Preparación algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "b4868e92-61e6-43f0-b644-2960ca8bbbc1",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "knn_6_i = KNeighborsClassifier(n_neighbors = 6)\n",
    "\n",
    "# 10-Fold Stratified Cross-Validation \n",
    "skf_i = StratifiedKFold(n_splits=10, random_state=None, shuffle=False)\n",
    "# random_state = None: el generador de números aleatorios utilizará un valor de semilla diferente \n",
    "    # cada vez que se ejecute el código, lo que resultará en un barajado diferente de los datos cada vez.\n",
    "# shuffle=False: los datos se utilizan en el orden del conjunto de datos de entrada y no se barajan.\n",
    "\n",
    "\n",
    "## Arrays para almacenar los resultados de cada iteración\n",
    "cm_list_i = []\n",
    "accu_list_i = []\n",
    "prec_list_i = []\n",
    "recall_list_i = []\n",
    "f1_list_i = []"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "06120d64-4d5f-4144-994e-a16d31a49e80",
   "metadata": {},
   "source": [
    "##### Ejecución algoritmo KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "1e609cfd-1b64-46c6-8fc3-5772ba69e569",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Iteración: 1\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 2\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 3\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 4\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 5\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 6\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 7\n",
      "Fold accuracy: 0.87\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 1]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.8666666666666667\n",
      "Precision: 0.8666666666666667\n",
      "Recall: 0.8666666666666667\n",
      "F1-Score: 0.8666666666666668\n",
      "******\n",
      "\n",
      "Iteración: 8\n",
      "Fold accuracy: 0.93\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 1 4]]\n",
      "\n",
      "Accuracy: 0.9333333333333333\n",
      "Precision: 0.9444444444444445\n",
      "Recall: 0.9333333333333332\n",
      "F1-Score: 0.9326599326599326\n",
      "******\n",
      "\n",
      "Iteración: 9\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n",
      "Iteración: 10\n",
      "Fold accuracy: 1.00\n",
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 5 0]\n",
      " [0 0 5]]\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1-Score: 1.0\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "n_iter_i = 0 ## Contador número de iteraciones del Cross-Validation\n",
    "\n",
    "# Evaluamos el modelo usando cross-validation\n",
    "for train_index_i, test_index_i in skf_i.split(X_scaled_iris, y_iris):\n",
    "    \n",
    "    n_iter_i +=1\n",
    "    print('Iteración:',n_iter_i)\n",
    "    \n",
    "    ## División del dataset en entrenamiento y test\n",
    "    X_train_i, X_test_i = X_scaled_iris[train_index_i], X_scaled_iris[test_index_i]\n",
    "    y_train_i, y_test_i = y_iris[train_index_i], y_iris[test_index_i]\n",
    "\n",
    "    ## Entrenar el algoritmo\n",
    "    knn_6_i.fit(X_train_i, y_train_i)\n",
    "    score_i = knn_6_i.score(X_test_i, y_test_i)\n",
    "    print(\"Fold accuracy: {:.2f}\".format(score_i))\n",
    "    \n",
    "    ## Predicción del algoritmo\n",
    "    y_pred_i = knn_6_i.predict(X_test_i)\n",
    "    \n",
    "    ## Métricas de evaluación\n",
    "    cm_i = confusion_matrix(y_test_i, y_pred_i)\n",
    "    accu_i = accuracy_score(y_test_i, y_pred_i)\n",
    "    prec_i = precision_score(y_test_i, y_pred_i, average='macro')\n",
    "    recall_i = recall_score(y_test_i, y_pred_i, average='macro')\n",
    "    f1_i = f1_score(y_test_i, y_pred_i, average='macro')\n",
    "    \n",
    "    ## Guardar resultados métricas de evaluación de cada iteración\n",
    "    cm_list_i.append(cm_i)\n",
    "    accu_list_i.append(accu_i)\n",
    "    prec_list_i.append(prec_i)\n",
    "    recall_list_i.append(recall_i)\n",
    "    f1_list_i.append(f1_i)\n",
    "    \n",
    "    ## Imprimir Métricas de evaluación en cada iteración\n",
    "    print('Matriz de Confusión:\\n',cm_i)\n",
    "    print('')\n",
    "    print('Accuracy:',accu_i)\n",
    "    print('Precision:',prec_i)\n",
    "    print('Recall:',recall_i)\n",
    "    print('F1-Score:',f1_i)\n",
    "    print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d4c828f3-ea9f-48ef-85aa-6df0ce030abe",
   "metadata": {},
   "source": [
    "##### Evaluación final del algoritmo KNN\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "245d3b0f-bacd-4bb3-802b-0aa93477e52a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Matriz de Confusión:\n",
      " [[5 0 0]\n",
      " [0 4 0]\n",
      " [0 0 4]]\n",
      "\n",
      "Accuracy: 0.967\n",
      "Precision: 0.97\n",
      "Recall: 0.967\n",
      "F1-Score: 0.966\n",
      "******\n",
      "\n"
     ]
    }
   ],
   "source": [
    "cm_mean_i = np.mean(cm_list_i, axis=0).astype(int)\n",
    "accu_mean_i = np.mean(accu_list_i, axis=0)\n",
    "prec_mean_i = np.mean(prec_list_i, axis=0)\n",
    "recall_mean_i = np.mean(recall_list_i, axis=0)\n",
    "f1_mean_i = np.mean(f1_list_i, axis=0)\n",
    "\n",
    "print('Matriz de Confusión:\\n',cm_mean_i)\n",
    "print('')\n",
    "print('Accuracy:',np.round(accu_mean_i,3))\n",
    "print('Precision:',np.round(prec_mean_i,3))\n",
    "print('Recall:',np.round(recall_mean_i,3))\n",
    "print('F1-Score:',np.round(f1_mean_i,3))\n",
    "print('******\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b05dce5-48f9-4add-82d1-ac8b4dd4e036",
   "metadata": {},
   "source": [
    "- Para k=6:\n",
    "    - Misma matriz de confusión que para k=2 y k=4.\n",
    "    - Observamos que los valores medios de la accuracy, precision, recall y F1-Score son ligeramente mejores que para k=2 y k=4. Por tanto, el rendimiento del modelo es mejor que para k=2 y k=4. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "235f452b-9320-4788-b7e6-802c53d23fb7",
   "metadata": {},
   "source": [
    "Conclusiones: \n",
    "\n",
    "- Para k=2, el clasificador alcanzó una accuracy de 0.953, lo que significa que clasificó correctamente el 95,3% de las muestras del conjunto de datos. La precision, el recall y el F1-score se sitúan en torno a 0.95.\n",
    "\n",
    "- Para k=4, la accuracy del clasificador descendió ligeramente a 0.947; la precision, el recall y el F1-score se sitúan en torno a 0.95.\n",
    "\n",
    "- Para k=6, la precisión del clasificador mejoró hasta 0.967; la precision, el recall y el F1-score  se sitúan en torno a 0.97.\n",
    "\n",
    "- En conclusión, con k=6 el modelo funciona mejor según las métricas anteriormente mencionadas debido al mayor número de vecinos considerados. Además, como los valores de k=2,4,6 están próximos entre sí, la diferencia de rendimiento también puede deberse a la variación aleatoria y al ruido de los datos. Por lo tanto, sería interesante probar con valores de k menos próximos entre sí.\n",
    "\n",
    "\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}