3-Evaluacion y supuestos OLS.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "39f8323e-35e6-49ff-9246-e6e6424198de",
   "metadata": {
    "tags": []
   },
   "source": [
    "# 3. Evaluación de un modelo OLS y supuestos a cumplir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "16932ce0-be93-451d-9ccd-d19270e7735f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt #para gráficos\n",
    "import statsmodels.api as sm #para modelos\n",
    "import pandas as pd\n",
    "from statsmodels.stats.outliers_influence import variance_inflation_factor #para calcular el VIF\n",
    "import scipy.stats as stats #para calcular los residuos estandarizados\n",
    "from sklearn.metrics import mean_squared_error #para calcular el MSE\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "9fe180b6-3989-4cdf-8957-46fc67d0eff6",
   "metadata": {
    "jupyter": {
     "source_hidden": true
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'python': '3.11.5 | packaged by Anaconda, Inc. | (main, Sep 11 2023, 13:26:23) [MSC v.1916 64 bit (AMD64)]',\n",
       " 'jupyterlab': '3.6.3',\n",
       " 'pandas': '2.0.3',\n",
       " 'numpy': '1.24.3',\n",
       " 'statsmodels': '0.14.0',\n",
       " 'matplotlib': '3.7.2',\n",
       " 'sklearn': '1.3.0',\n",
       " 'scipy': '1.11.1'}"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import matplotlib\n",
    "import statsmodels\n",
    "import scipy\n",
    "import sklearn\n",
    "import sys\n",
    "\n",
    "# Function to get the version of JupyterLab\n",
    "def get_jupyterlab_version():\n",
    "    try:\n",
    "        import jupyterlab\n",
    "        return jupyterlab.__version__\n",
    "    except ImportError:\n",
    "        return 'Not Installed'\n",
    "    \n",
    "library_versions = {\n",
    "    \"python\": sys.version,\n",
    "    \"jupyterlab\": get_jupyterlab_version(),\n",
    "    \"pandas\": pd.__version__,\n",
    "    \"numpy\": np.__version__,\n",
    "    \"statsmodels\": sm.__version__,\n",
    "    \"matplotlib\": matplotlib.__version__,\n",
    "    \"sklearn\": sklearn.__version__,  # Add this line\n",
    "    \"scipy\": scipy.__version__,\n",
    "}\n",
    "\n",
    "library_versions"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48d1041f-e14b-43d8-acc3-aff948b51927",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 3.1. Cargamos el modelo obtenido"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "74477682-64ab-4477-bf17-4512ea227880",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "mod_antiguo = sm.load('modelos/clases/modeloOLS_c1.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "1525af8c-eda0-467f-bd06-fdc542720997",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                            OLS Regression Results                            \n",
      "==============================================================================\n",
      "Dep. Variable:              SalePrice   R-squared:                       0.674\n",
      "Model:                            OLS   Adj. R-squared:                  0.672\n",
      "Method:                 Least Squares   F-statistic:                     358.8\n",
      "Date:                Wed, 13 Mar 2024   Prob (F-statistic):               0.00\n",
      "Time:                        17:17:21   Log-Likelihood:                -16575.\n",
      "No. Observations:                1399   AIC:                         3.317e+04\n",
      "Df Residuals:                    1390   BIC:                         3.322e+04\n",
      "Df Model:                           8                                         \n",
      "Covariance Type:            nonrobust                                         \n",
      "================================================================================\n",
      "                   coef    std err          t      P>|t|      [0.025      0.975]\n",
      "--------------------------------------------------------------------------------\n",
      "const        -9.912e+05   1.38e+06     -0.719      0.472   -3.69e+06    1.71e+06\n",
      "YearBuilt      839.0413     40.215     20.864      0.000     760.153     917.930\n",
      "2ndFlrSF       -20.4224      3.781     -5.401      0.000     -27.840     -13.005\n",
      "GrLivArea       89.7873      3.863     23.244      0.000      82.210      97.365\n",
      "FullBath      3397.8437   2633.955      1.290      0.197   -1769.112    8564.799\n",
      "HalfBath      -950.0014   2634.194     -0.361      0.718   -6117.426    4217.424\n",
      "BedroomAbvGr -8098.6069   1616.196     -5.011      0.000   -1.13e+04   -4928.160\n",
      "TotRmsAbvGrd   369.5097   1202.816      0.307      0.759   -1990.022    2729.041\n",
      "YrSold        -299.3928    684.212     -0.438      0.662   -1641.592    1042.806\n",
      "==============================================================================\n",
      "Omnibus:                      585.271   Durbin-Watson:                   2.021\n",
      "Prob(Omnibus):                  0.000   Jarque-Bera (JB):            20326.648\n",
      "Skew:                          -1.279   Prob(JB):                         0.00\n",
      "Kurtosis:                      21.498   Cond. No.                     4.87e+06\n",
      "==============================================================================\n",
      "\n",
      "Notes:\n",
      "[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.\n",
      "[2] The condition number is large, 4.87e+06. This might indicate that there are\n",
      "strong multicollinearity or other numerical problems.\n"
     ]
    }
   ],
   "source": [
    "print(mod_antiguo.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e291eeb0-4945-4a40-be31-5a5221d9cb78",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 3.2. Capacidad de explicación e impacto de variables"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9ded1825-ea1c-412a-b341-d115a23b4fb5",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.2.1. Raíz del Error Cuadrático Medio (RMSE)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96853500-9405-4950-b271-5db71e997431",
   "metadata": {},
   "source": [
    "El RMSE es una medida de la diferencia promedio entre los valores reales y las predicciones del modelo, en la misma unidad que los datos originales. Cuanto menor sea el valor del RMSE, mejor será la capacidad predictiva del modelo."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "59191945-5c4d-4ddd-a244-6b113e0e3bb1",
   "metadata": {},
   "source": [
    "Primero, importemos nuestros datos."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "034e59d7-f276-43c5-b3ee-e7d33aec6da3",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 1399 entries, 0 to 1398\n",
      "Data columns (total 9 columns):\n",
      " #   Column        Non-Null Count  Dtype\n",
      "---  ------        --------------  -----\n",
      " 0   YearBuilt     1399 non-null   int64\n",
      " 1   2ndFlrSF      1399 non-null   int64\n",
      " 2   GrLivArea     1399 non-null   int64\n",
      " 3   FullBath      1399 non-null   int64\n",
      " 4   HalfBath      1399 non-null   int64\n",
      " 5   BedroomAbvGr  1399 non-null   int64\n",
      " 6   TotRmsAbvGrd  1399 non-null   int64\n",
      " 7   YrSold        1399 non-null   int64\n",
      " 8   SalePrice     1399 non-null   int64\n",
      "dtypes: int64(9)\n",
      "memory usage: 98.5 KB\n"
     ]
    }
   ],
   "source": [
    "#Importo el archivo que usé\n",
    "bdd_usada=pd.read_csv('data/clases/HousingPrices_Filtered_C1.csv')\n",
    "bdd_usada.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b62ef475-aadd-4b4c-9d1f-fa52bc009701",
   "metadata": {},
   "source": [
    "En el cuaderno anterior, si al guardar mi archivo filtrado no hubiera escrito `index=False`, ahora tendría una columna llamada `Unnamed: 0` en mi *dataframe*. Y tendríamos que ponernos a editar el *dataframe* para eliminarlo."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a29071c3-7027-4d72-911c-00fa4b1afbb4",
   "metadata": {},
   "source": [
    "¿Qué creen que necesitamos para encontrar los errores? Si ya teníamos nuestro modelo con una constante y un coeficiente, sólo nos faltaría aplicarlo a nuestros valores reales del *dataframe* para compararlos."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "247f57d9-684c-4a77-9147-a367b33e7385",
   "metadata": {},
   "source": [
    "Empecemos creando objetos para indicarle a Python cuáles son mis *X* y mis *Y*."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "c3ba9b04-52ca-47af-98a1-bbc418f3c5c0",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "X=bdd_usada.drop(['SalePrice'],axis=1)\n",
    "y=bdd_usada['SalePrice']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2e322cfa-6282-4885-9da9-a8dfb19c18f8",
   "metadata": {},
   "source": [
    "A mis *X* les añado una constante. Recuerden que nuestro modelo tiene una."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5154c1b0-4075-44a8-b8fe-4b662aae865d",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "X = sm.add_constant(X) #es una función de la libería 'sm'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "740af05e-63c3-4e49-9c3b-a5e05c7269e9",
   "metadata": {
    "tags": []
   },
   "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>const</th>\n",
       "      <th>YearBuilt</th>\n",
       "      <th>2ndFlrSF</th>\n",
       "      <th>GrLivArea</th>\n",
       "      <th>FullBath</th>\n",
       "      <th>HalfBath</th>\n",
       "      <th>BedroomAbvGr</th>\n",
       "      <th>TotRmsAbvGrd</th>\n",
       "      <th>YrSold</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2003</td>\n",
       "      <td>854</td>\n",
       "      <td>1710</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>2008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1976</td>\n",
       "      <td>0</td>\n",
       "      <td>1262</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>2007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2001</td>\n",
       "      <td>866</td>\n",
       "      <td>1786</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>2008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1915</td>\n",
       "      <td>756</td>\n",
       "      <td>1717</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>2006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0</td>\n",
       "      <td>2000</td>\n",
       "      <td>1053</td>\n",
       "      <td>2198</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>2008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1394</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1999</td>\n",
       "      <td>694</td>\n",
       "      <td>1647</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>2007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1395</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1978</td>\n",
       "      <td>0</td>\n",
       "      <td>2073</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1396</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1941</td>\n",
       "      <td>1152</td>\n",
       "      <td>2340</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>9</td>\n",
       "      <td>2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1397</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1950</td>\n",
       "      <td>0</td>\n",
       "      <td>1078</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>2010</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1398</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1965</td>\n",
       "      <td>0</td>\n",
       "      <td>1256</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>2008</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1399 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      const  YearBuilt  2ndFlrSF  GrLivArea  FullBath  HalfBath  BedroomAbvGr  \\\n",
       "0       1.0       2003       854       1710         2         1             3   \n",
       "1       1.0       1976         0       1262         2         0             3   \n",
       "2       1.0       2001       866       1786         2         1             3   \n",
       "3       1.0       1915       756       1717         1         0             3   \n",
       "4       1.0       2000      1053       2198         2         1             4   \n",
       "...     ...        ...       ...        ...       ...       ...           ...   \n",
       "1394    1.0       1999       694       1647         2         1             3   \n",
       "1395    1.0       1978         0       2073         2         0             3   \n",
       "1396    1.0       1941      1152       2340         2         0             4   \n",
       "1397    1.0       1950         0       1078         1         0             2   \n",
       "1398    1.0       1965         0       1256         1         1             3   \n",
       "\n",
       "      TotRmsAbvGrd  YrSold  \n",
       "0                8    2008  \n",
       "1                6    2007  \n",
       "2                6    2008  \n",
       "3                7    2006  \n",
       "4                9    2008  \n",
       "...            ...     ...  \n",
       "1394             7    2007  \n",
       "1395             7    2010  \n",
       "1396             9    2010  \n",
       "1397             5    2010  \n",
       "1398             6    2008  \n",
       "\n",
       "[1399 rows x 9 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Reviso que esté en orden\n",
    "X"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7939b8b5-cc44-461b-b43a-df90f86427ec",
   "metadata": {},
   "source": [
    "Ahora sí calculo el RMSE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "02520bb9-4faa-4c16-941e-96c29d613b40",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "33821.01127443376"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Hacer predicciones con el modelo\n",
    "predicciones = mod_antiguo.predict(X)\n",
    "\n",
    "# Calcular el MSE\n",
    "mse = mean_squared_error(y, predicciones) #mean_squared_error ya es una función predefinida de 'sm'\n",
    "\n",
    "#Ahora el RMSE\n",
    "rmse = np.sqrt(mse) #usamos numpy (cargada como 'np' al inicio) para sacar la raíz cuadrada\n",
    "\n",
    "#Le pido que me muestre cuál es el valor\n",
    "rmse "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5dedde11-aaef-4663-92a7-a06bbed98d1c",
   "metadata": {},
   "source": [
    "En este caso, podemos decir que **estamos alejados US$ 33.821 del valor real** de las viviendas de nuestros datos."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5d1a8133-4afa-4d21-a9ff-1edcff5d4c5d",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.2.2. $R^2$"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "86b56349-2d3b-46d0-b98c-04a19bdd0ecf",
   "metadata": {
    "tags": []
   },
   "source": [
    "El $R^2$ es la medida estadística de **qué tan cerca están los datos de la línea de regresión**. También se conoce como coeficiente de determinación, mide cuánta varianza en la variable Y es explicada por todas nuestras X.\n",
    "\n",
    "En términos sencillos, **qué porcentaje de Y nos permite explicar nuestro modelo**. Cuando estamos tratando de estimar o predecir un valor, es un valor importante.\n",
    "\n",
    "En nuestro modelo, su valor es de 67.4%."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0062c4a3-2a5e-4954-9d96-2da9aba19718",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.2.3. Los p-valores"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "30dc07e8-9292-4c16-bf59-0755c5fed4dc",
   "metadata": {},
   "source": [
    "Es lo que aparece en nuestra tabla como `P>|t|`. Se refiere a la significancia estadística de los coeficientes de las variables X. \n",
    "\n",
    "Dicho de otro modo, nos indican qué tan confiables son esos coeficientes, y, por tanto, esas variables. Se asume que **si su valor es menor a 0.05, el coeficiente es estadísticamente significativo**. Cuando se hacen modelos de precios hedónicos, se prefiere que sólo se tomen en cuenta variables que cumplen con esta característica.\n",
    "\n",
    "En nuestro modelo, `FullBath`, `HalfBath`, `TotRmsAbvGrd` y `YrSold` no tienen coeficientes signifativos, por lo que no podemos 'confiar' en ellas."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "97ad1b67-21d0-42c4-985b-72baeb04da69",
   "metadata": {
    "tags": []
   },
   "source": [
    "## 3.3. Supuestos a cumplir"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8f342710-798e-4300-a95d-a13ba70b0cfb",
   "metadata": {},
   "source": [
    "Más allá de qué tan bueno sea un modelo a partir de lo revisado en el apartado anterior, es necesario que estén muy seguros/as de que cumpla con algunos supuestos.\n",
    "\n",
    "Como les comenté al inicio, hemos elaborado un modelo de regresión OLS. Este tipo de modelos requieren que se cumpla con lo siguiente:"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "78288db8-78fb-4017-bf13-c46b39f56956",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.3.1. Linealidad de los datos"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3de92874-ffff-40fe-b51d-d7107bdc3e2a",
   "metadata": {},
   "source": [
    "Se espera una relación lineal $X$ e $y$. Esto es sencillo de ver a través de un gráfico de dispersión cuando tenemos una regresión lineal simple (una $X$ y una $y$), pero en el caso de una regresión lineal múltiple usualmente se utilizan los *gráficos de residuos*. Idealmente, en este gráfico no se va a observar ningún patrón claro de relación entre el eje vertical y horizontal. Si no fuese así, **se recomienda transformar las variables** ($\\log(X)$, $\\sqrt{X}$, $X^2$).\n",
    "\n",
    "Esta transformación suele hacerse primero en la variable a estimar ($y$), y si aún así el modelo sigue presentando errores, se puede hacer en las variables independientes. Ten en cuenta que eso implicará una [interpretación distinta de los coeficientes del modelo](https://economipedia.com/definiciones/logaritmos-en-econometria.html)."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "63127977-2c89-4d34-bc5d-3b3045a456e2",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.3.2. Homosedasticidad"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e0fbdc3-3a69-4e57-b03a-207a988f7f30",
   "metadata": {},
   "source": [
    "En términos sencillos, la homocedasticidad significa que la variabilidad de los errores (o residuos) de un modelo estadístico es constante en todas las posibles combinaciones de valores de las variables independientes.\n",
    "\n",
    "Si estuviéramos construyendo un modelo de regresión para predecir el precio de pisos basado únicamente en su superficie, la homocedasticidad implicaría que, sin importar el tamaño de la casa, la variabilidad en los errores del modelo es constante. En otras palabras, los errores no se vuelven más grandes ni más pequeños a medida que cambia el tamaño de la casa.\n",
    "\n",
    "Esto también se puede observar con el *gráfico de residuos*: si los residuos están distribuidos uniformemente alrededor de la línea horizontal en $y$=0, esto sugiere que la varianza de los errores es constante a lo largo de los valores ajustados.\n",
    "\n",
    "Tengan en cuenta que la homosedasticidad (aquello que **sí buscamos**) es lo contrario de la heterosedasticidad."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "af18d57e-08c0-4d3c-b722-4270ea4c4b7c",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calcular los residuos del modelo\n",
    "residuos = mod_antiguo.resid ## 'mod_antiguo' es como llamé a mi modelo al cargarlo ##\n",
    "\n",
    "# Graficar los residuos\n",
    "plt.figure(figsize=(8, 6))\n",
    "plt.scatter(mod_antiguo.fittedvalues, residuos) ## revisa 'mod_antiguo' ##\n",
    "plt.axhline(y=0, color='r', linestyle='--')\n",
    "plt.title('Gráfico de residuos')\n",
    "plt.xlabel('Valores ajustados')\n",
    "plt.ylabel('Residuos')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5b64643-c7d7-4b83-b7a1-d137a9e57771",
   "metadata": {
    "tags": []
   },
   "source": [
    "#### 3.3.2.1. Detección de outliers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd7062ec-b080-41e0-9260-0a81826d005b",
   "metadata": {
    "tags": []
   },
   "source": [
    "Si bien los resultados del gráfico nos indican que tenemos linealidad en los datos y también cumplimos con el supuesto de homosedasticidad, notamos que existen dos outliers: aquellos que tienen un error menor de US$ -250.000. Si bien podríamos pensar que al ser sólo dos no tienen un gran impacto en nuestro modelo, sí \"fastidian\" nuestro error. Por tanto, idealmente deberíamos quitarlos y volver a correr el modelo."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "04db9cf8-c5a7-4e4d-9a44-7bc568c5771b",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        -318.565692\n",
       "1       17609.123625\n",
       "2       10519.772420\n",
       "3        4605.722838\n",
       "4       11675.527163\n",
       "            ...     \n",
       "1455   -28003.273097\n",
       "1456   -27857.770227\n",
       "1457    66599.772302\n",
       "1458    13136.979924\n",
       "1459    -1975.460941\n",
       "Length: 1399, dtype: float64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Noten que llamé 'residuos' al objeto que \n",
    "#contiene los residuos de mi modelo en la celda anterior \n",
    "residuos"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "92cc0604-2b28-41f0-ba9d-8f2ae2c5eea3",
   "metadata": {},
   "source": [
    "¡Noten para qué nos sirve el índice! Si ahora filtramos aquellos valores por debajo de US$ -250.000, podemos saber qué filas son los outliers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3cee15ef-7f07-403c-a722-cf43277a2a42",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(498, -284570.61729873205), (1241, -405574.82405305805)]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "outliers2= [(indice, valor) for \n",
    "            indice, valor in enumerate(residuos) ## revisar 'residuos' ##\n",
    "            if valor < -250000]\n",
    "\n",
    "outliers2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9c9644f9-6a49-49d9-876f-c32fc1d90e14",
   "metadata": {
    "tags": []
   },
   "source": [
    "Revisamos el *dataframe* que usamos para el modelo. Ya lo habíamos cargado en los pasos anteriores de este cuaderno."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "0a85ffa2-664a-43bf-bbb9-aeac7fda92be",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 1399 entries, 0 to 1398\n",
      "Data columns (total 9 columns):\n",
      " #   Column        Non-Null Count  Dtype\n",
      "---  ------        --------------  -----\n",
      " 0   YearBuilt     1399 non-null   int64\n",
      " 1   2ndFlrSF      1399 non-null   int64\n",
      " 2   GrLivArea     1399 non-null   int64\n",
      " 3   FullBath      1399 non-null   int64\n",
      " 4   HalfBath      1399 non-null   int64\n",
      " 5   BedroomAbvGr  1399 non-null   int64\n",
      " 6   TotRmsAbvGrd  1399 non-null   int64\n",
      " 7   YrSold        1399 non-null   int64\n",
      " 8   SalePrice     1399 non-null   int64\n",
      "dtypes: int64(9)\n",
      "memory usage: 98.5 KB\n"
     ]
    }
   ],
   "source": [
    "bdd_usada.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b0a32e88-6dd4-45a5-b35a-b40061bc7855",
   "metadata": {},
   "source": [
    "Noten que tenemos 1399 filas, con un índice que va desde 0 a 1398. Quitemos de una vez esos dos outliers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "7c46478e-8131-441e-8b66-75eb80ed2e10",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 1397 entries, 0 to 1398\n",
      "Data columns (total 9 columns):\n",
      " #   Column        Non-Null Count  Dtype\n",
      "---  ------        --------------  -----\n",
      " 0   YearBuilt     1397 non-null   int64\n",
      " 1   2ndFlrSF      1397 non-null   int64\n",
      " 2   GrLivArea     1397 non-null   int64\n",
      " 3   FullBath      1397 non-null   int64\n",
      " 4   HalfBath      1397 non-null   int64\n",
      " 5   BedroomAbvGr  1397 non-null   int64\n",
      " 6   TotRmsAbvGrd  1397 non-null   int64\n",
      " 7   YrSold        1397 non-null   int64\n",
      " 8   SalePrice     1397 non-null   int64\n",
      "dtypes: int64(9)\n",
      "memory usage: 109.1 KB\n"
     ]
    }
   ],
   "source": [
    "nueva_bdd = bdd_usada.drop([498,1241], axis=0) ## revisar 'nueva_bdd' ## \n",
    "                                               #axis 0 son filas\n",
    "nueva_bdd.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "53dce5e5-64be-4555-a6d3-9021a8bbb8b5",
   "metadata": {},
   "source": [
    "A partir de aquí podríamos guardar la nueva base de datos y volver a correr el modelo."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79401de3-f280-468c-bbf9-5bde1abe2f0a",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.3.3. Normalidad de los errores"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "330f8be5-811d-4024-b40b-2b53346b0fd2",
   "metadata": {},
   "source": [
    "Suponer que los errores tienen una distribución normal ayuda a que el modelo se ajuste bien a los datos reales. Esto significa que los valores reales tienden a estar cerca de los valores predichos por el modelo en la mayoría de los casos."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "88699506-3253-464b-b0b7-846eb2b1f373",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8, 6))\n",
    "plt.hist(residuos, ## revisar 'residuos' ## \n",
    "         bins=30, density=True, alpha=0.6, color='b')\n",
    "plt.title('Histograma de los errores')\n",
    "plt.xlabel('Errores')\n",
    "plt.ylabel('Densidad')\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "704fe00f-6847-4609-8cf7-cca174f9afe3",
   "metadata": {},
   "source": [
    "El gráfico nos muestra una distribución normal, pero con datos que lo distorsionan hacia la izquierda. ¿Adivinen cuáles son? Los outliers que encontramos hace un rato.\n",
    "\n",
    "Usualmente, se suele complementar con la generación del siguiente gráfico:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "ecbc70f1-ccda-460b-97fa-6d857cd42830",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calcular los residuos estandarizados\n",
    "residuos_estandarizados = mod_antiguo.get_influence().resid_studentized_internal ## revisar 'mod_antiguo' ##\n",
    "\n",
    "# Calcular la distribución acumulativa de los residuos estandarizados\n",
    "distribucion_acumulativa = stats.probplot(residuos_estandarizados, dist=\"norm\", plot=plt)\n",
    "\n",
    "# Personalizar el gráfico\n",
    "plt.title('Gráfico P-P normal de residuos estandarizados')\n",
    "plt.xlabel('Cuantiles teóricos')\n",
    "plt.ylabel('Cuantiles de los residuos estandarizados')\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f8aee69-58d6-4879-888e-ff6e39e0da72",
   "metadata": {},
   "source": [
    "Si los puntos en el gráfico se aproximan a una línea diagonal, eso sugiere que los residuos siguen una distribución normal. Si hay desviaciones significativas de la línea diagonal, eso podría indicar que los residuos no siguen una distribución normal y podría requerir una investigación adicional."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d5d82a51-7bb4-4801-8425-fcc976821f9e",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.3.4. Multicolinealidad"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d7686fb-4675-4c4f-8c4f-1929975ec336",
   "metadata": {},
   "source": [
    "La presencia de multicolinealidad puede hacer difícil separar los efectos individuales de las variables colineales en la respuesta. Esto puede engañar tanto al modelo (influenciando en el p-valor de las variables) como a nosotros al momento de interpretar los resultados.\n",
    "\n",
    "Nosotros ya hemos hecho un análisis de correlación y hemos visto que la mayoría de nuestras variables se relacionan significativamente entre ellas. Sin embargo, hay un punto hasta el que suele ser permitido ese nivel de correlación. Eso se mide a través del Factor de Inflación de la Varianza (VIF, por sus siglas en inglés). Si hay valores por encima de 3, se asume que esas variables no son recomendables de tomar en cuenta.\n",
    "\n",
    "Ojo: este valor varía mucho entre investigaciones. Algunas consideran 3, otras 10, otras 40. Consulten bibliografía reciente sobre el tema que están tratando para tener un estándar (o consulten con Carlos)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "64d943cc-962c-4b4b-af8e-756019b5ccb1",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "       Variable           VIF\n",
      "0         const  2.306952e+06\n",
      "1     YearBuilt  1.768579e+00\n",
      "2      2ndFlrSF  3.083363e+00\n",
      "3     GrLivArea  4.184106e+00\n",
      "4      FullBath  2.431363e+00\n",
      "5      HalfBath  2.110762e+00\n",
      "6  BedroomAbvGr  2.060995e+00\n",
      "7  TotRmsAbvGrd  4.192325e+00\n",
      "8        YrSold  1.002196e+00\n"
     ]
    }
   ],
   "source": [
    "#Al inicio de este cuaderno, importé todas mis X en una variable llamada 'X'\n",
    "\n",
    "# Calcular el VIF para cada variable independiente\n",
    "vif = pd.DataFrame()\n",
    "vif[\"Variable\"] = X.columns ## revisar 'X' ##\n",
    "vif[\"VIF\"] = [variance_inflation_factor(X.values, i) ## revisar 'X' ##\n",
    "              for i in range(X.shape[1])] ## revisar 'X' ##\n",
    "\n",
    "# Imprimir el resultado\n",
    "print(vif)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2aa5f4be-e009-4739-90f8-a2b42f9e1346",
   "metadata": {},
   "source": [
    "Vemos que tanto `GrLivArea` como `TotRmsAbvGrd` tienen un VIF más alto de lo esperado, probablemente porque su correlación es muy alta. A partir de aquí pueden ir eliminando aquellas que les parecen menos relevantes, volver a correr un modelo y medir el VIF para ver si cumplen con un valor menor al esperado."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "475b133f-c0d1-4fa9-a68c-0793627950c8",
   "metadata": {
    "tags": []
   },
   "source": [
    "### 3.3.5. Autocorrelación de errores"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12b12bc0-16af-4d0e-9921-7412ba250f46",
   "metadata": {},
   "source": [
    "De manera ideal, los errores de cada fila son independientes unos de otros. Cuando eso no sucede, hablamos de una autocorrelación de los errores o residuos del modelo. \n",
    "\n",
    "La autocorrelación es un fenómeno que se presenta en muestras que contengan de datos asociados al tiempo, aunque también puede presentarse cuando se trabaja con datos de corte transversal, usualmente cuando están georreferenciados (lo que se conoce como *autocorrelación espacial*).\n",
    "\n",
    "Además, puede presentarse porque nuestra variable a estimar (*Y*) presenta un comportamiento cíclico que no viene explicado por las variables explicativas (*X*), si no que dicho comportamiento cíclico está recogido en el término de error del modelo. Esto nos habla de que nuestro modelo es malo: le falta algo, una variable que no hemos sido capaces de introducir en el modelo (¡y que no sabemos cuál es!).\n",
    "\n",
    "Para medir esta autocorrelación de los errores, hay una medida llamada **Durbin-Watson**, que aparece en el resumen de nuestro modelo. **Si su valor está entre 1.5 y 2.5, se entiende que no existe una autocorrelación de los errores del modelo**."
   ]
  }
 ],
 "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}