cremerfPreds.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "id": "3d08028c-be52-4d84-a995-34b690c3e8a5",
      "metadata": {
        "id": "3d08028c-be52-4d84-a995-34b690c3e8a5"
      },
      "source": [
        "# Making Estimations with our NBA dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4ff2f3cb-7e1a-4705-bce7-44a56c0113e1",
      "metadata": {
        "id": "4ff2f3cb-7e1a-4705-bce7-44a56c0113e1"
      },
      "source": [
        "Let's explore how to make predictive models for regression and classification tasks using ScikitLearn with our extracted, preprocess and clean NBA data."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "5b732b5b-8fcc-4917-85a1-035bfc00a7dd",
      "metadata": {
        "id": "5b732b5b-8fcc-4917-85a1-035bfc00a7dd"
      },
      "source": [
        "#### The goals of this project are:\n",
        "\n",
        "- Define a prediction task\n",
        "- Selecting evaluation metrics and baseline models\n",
        "- Perform feature engineering and standarization\n",
        "- Training and using predictive models: Univariable and Multivariate Linear Regression, Classification\n",
        "- Understand how Gradient Descent works by implementing a Linear Regressor in python"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "73bd1212-887d-4b10-a277-30156aaf410b",
      "metadata": {
        "id": "73bd1212-887d-4b10-a277-30156aaf410b",
        "tags": []
      },
      "source": [
        "### 1. Regression Models to Estimate players Salaries"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e51670c9-30ec-4dcf-b3f4-e5ba90019818",
      "metadata": {
        "id": "e51670c9-30ec-4dcf-b3f4-e5ba90019818"
      },
      "source": [
        "A player's salary can be explained by a lot of factors: \n",
        "\n",
        "- Years in the league: NBA contracts are limited in price, players have caps on how much money they make depending on their years playing in the league\n",
        "- Age: Highly correlated to the previous one, NBA players have a minimun age of 19 years\n",
        "- Draft position: players out of USA universities are selected each year through a process called draft, in which each team selects a player in a predefined order for 2 rounds (meaning only 60 players are drafted each year). A rookie contract value depends on the draft position. (Undrafted players have to negotiate a salary independently)\n",
        "- Performance in statistical categories: after the rookie contract, players can negotiate freely with teams, and usually statistical categories like points, rebounds and assists are extremely important.  \n",
        "- Team: Even though there are limits to what all teams can expend in players salary, they all allocate their resources differently, and can be more willing to pay the penalties incurred by exceeding those limitations.\n",
        "- Year in which the contract was signed: the mininum and maximum values of contracts possibles in the NBA changes depending on league revenue (television contracts, sponsorships, etc), so if the amount of money the teams make increases, players can negotiate bigger contracts.\n",
        "- Intangibles: players can be valued also in things that are not easily translated to a statistic, good defensive players might not always have great stats but can be extremely important for a team, the same can be said about veteran players that help young ones to learn and are good locker-room guys."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "6f110fec-8919-4427-89f7-b3e581176703",
      "metadata": {
        "id": "6f110fec-8919-4427-89f7-b3e581176703"
      },
      "source": [
        "### Analyzing our data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "0315863d",
      "metadata": {},
      "outputs": [],
      "source": [
        "!pip install -r requirements.txt"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "id": "18653712",
      "metadata": {},
      "outputs": [],
      "source": [
        "import pylab \n",
        "import datetime\n",
        "import warnings\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import scipy.stats as stats\n",
        "import matplotlib.pyplot as plt\n",
        "import sklearn.model_selection as model_selection\n",
        "from sklearn.metrics import mean_absolute_error\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn import linear_model\n",
        "import math\n",
        "from sklearn.tree import DecisionTreeRegressor\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "id": "bdac389b-e5f8-4563-a187-36b016c0641c",
      "metadata": {
        "id": "bdac389b-e5f8-4563-a187-36b016c0641c"
      },
      "outputs": [],
      "source": [
        "data = pd.read_csv(\"project2_players_df.csv\")\n",
        "data.dropna(subset=[\"PTS\"], inplace=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e098b76d-c520-474b-9998-a263354fe27b",
      "metadata": {
        "id": "e098b76d-c520-474b-9998-a263354fe27b"
      },
      "source": [
        "All players statistics (\"PTS\", \"REB\", \"AST\",\"STL\",\"BLK\") against salary"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "id": "50d589f1-25c1-48f8-88f3-e2c7f0e72625",
      "metadata": {
        "id": "50d589f1-25c1-48f8-88f3-e2c7f0e72625"
      },
      "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>PERSON_ID</th>\n",
              "      <th>PLAYER_NAME</th>\n",
              "      <th>TEAM_NAME</th>\n",
              "      <th>POSITION</th>\n",
              "      <th>HEIGHT</th>\n",
              "      <th>WEIGHT</th>\n",
              "      <th>COUNTRY</th>\n",
              "      <th>BIRTHDATE</th>\n",
              "      <th>SEASON_EXP</th>\n",
              "      <th>DRAFT_NUMBER</th>\n",
              "      <th>GP</th>\n",
              "      <th>MIN</th>\n",
              "      <th>PTS</th>\n",
              "      <th>REB</th>\n",
              "      <th>AST</th>\n",
              "      <th>STL</th>\n",
              "      <th>BLK</th>\n",
              "      <th>GAME_DATE</th>\n",
              "      <th>Salary</th>\n",
              "      <th>AGE</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1630173</td>\n",
              "      <td>Precious Achiuwa</td>\n",
              "      <td>Raptors</td>\n",
              "      <td>Forward</td>\n",
              "      <td>203</td>\n",
              "      <td>102</td>\n",
              "      <td>Nigeria</td>\n",
              "      <td>1999-09-19</td>\n",
              "      <td>1</td>\n",
              "      <td>20</td>\n",
              "      <td>115.0</td>\n",
              "      <td>17.3</td>\n",
              "      <td>6.5</td>\n",
              "      <td>5.1</td>\n",
              "      <td>0.8</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.5</td>\n",
              "      <td>2022-03-06</td>\n",
              "      <td>2711280</td>\n",
              "      <td>22 years, 5 months, 25 days</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>203500</td>\n",
              "      <td>Steven Adams</td>\n",
              "      <td>Grizzlies</td>\n",
              "      <td>Center</td>\n",
              "      <td>211</td>\n",
              "      <td>120</td>\n",
              "      <td>New Zealand</td>\n",
              "      <td>1993-07-20</td>\n",
              "      <td>8</td>\n",
              "      <td>12</td>\n",
              "      <td>648.0</td>\n",
              "      <td>26.8</td>\n",
              "      <td>9.3</td>\n",
              "      <td>7.9</td>\n",
              "      <td>1.4</td>\n",
              "      <td>0.9</td>\n",
              "      <td>0.9</td>\n",
              "      <td>2022-03-06</td>\n",
              "      <td>17073171</td>\n",
              "      <td>28 years, 7 months, 24 days</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1628389</td>\n",
              "      <td>Bam Adebayo</td>\n",
              "      <td>Heat</td>\n",
              "      <td>Center</td>\n",
              "      <td>206</td>\n",
              "      <td>116</td>\n",
              "      <td>USA</td>\n",
              "      <td>1997-07-18</td>\n",
              "      <td>4</td>\n",
              "      <td>14</td>\n",
              "      <td>326.0</td>\n",
              "      <td>28.0</td>\n",
              "      <td>13.2</td>\n",
              "      <td>8.2</td>\n",
              "      <td>3.5</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.9</td>\n",
              "      <td>2022-03-07</td>\n",
              "      <td>28103550</td>\n",
              "      <td>24 years, 7 months, 26 days</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1630583</td>\n",
              "      <td>Santi Aldama</td>\n",
              "      <td>Grizzlies</td>\n",
              "      <td>Forward</td>\n",
              "      <td>211</td>\n",
              "      <td>98</td>\n",
              "      <td>Spain</td>\n",
              "      <td>2001-01-10</td>\n",
              "      <td>0</td>\n",
              "      <td>30</td>\n",
              "      <td>27.0</td>\n",
              "      <td>10.2</td>\n",
              "      <td>3.3</td>\n",
              "      <td>2.4</td>\n",
              "      <td>0.5</td>\n",
              "      <td>0.1</td>\n",
              "      <td>0.3</td>\n",
              "      <td>2022-03-06</td>\n",
              "      <td>1994520</td>\n",
              "      <td>21 years, 2 months, 6 days</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>200746</td>\n",
              "      <td>LaMarcus Aldridge</td>\n",
              "      <td>Nets</td>\n",
              "      <td>Center</td>\n",
              "      <td>211</td>\n",
              "      <td>113</td>\n",
              "      <td>USA</td>\n",
              "      <td>1985-07-19</td>\n",
              "      <td>15</td>\n",
              "      <td>2</td>\n",
              "      <td>1073.0</td>\n",
              "      <td>33.7</td>\n",
              "      <td>19.1</td>\n",
              "      <td>8.1</td>\n",
              "      <td>1.9</td>\n",
              "      <td>0.7</td>\n",
              "      <td>1.1</td>\n",
              "      <td>2022-03-06</td>\n",
              "      <td>2641691</td>\n",
              "      <td>36 years, 7 months, 25 days</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   PERSON_ID        PLAYER_NAME  TEAM_NAME POSITION  HEIGHT  WEIGHT  \\\n",
              "0    1630173   Precious Achiuwa    Raptors  Forward     203     102   \n",
              "1     203500       Steven Adams  Grizzlies   Center     211     120   \n",
              "2    1628389        Bam Adebayo       Heat   Center     206     116   \n",
              "3    1630583       Santi Aldama  Grizzlies  Forward     211      98   \n",
              "4     200746  LaMarcus Aldridge       Nets   Center     211     113   \n",
              "\n",
              "       COUNTRY   BIRTHDATE  SEASON_EXP DRAFT_NUMBER      GP   MIN   PTS  REB  \\\n",
              "0      Nigeria  1999-09-19           1           20   115.0  17.3   6.5  5.1   \n",
              "1  New Zealand  1993-07-20           8           12   648.0  26.8   9.3  7.9   \n",
              "2          USA  1997-07-18           4           14   326.0  28.0  13.2  8.2   \n",
              "3        Spain  2001-01-10           0           30    27.0  10.2   3.3  2.4   \n",
              "4          USA  1985-07-19          15            2  1073.0  33.7  19.1  8.1   \n",
              "\n",
              "   AST  STL  BLK   GAME_DATE    Salary                          AGE  \n",
              "0  0.8  0.4  0.5  2022-03-06   2711280  22 years, 5 months, 25 days  \n",
              "1  1.4  0.9  0.9  2022-03-06  17073171  28 years, 7 months, 24 days  \n",
              "2  3.5  1.0  0.9  2022-03-07  28103550  24 years, 7 months, 26 days  \n",
              "3  0.5  0.1  0.3  2022-03-06   1994520   21 years, 2 months, 6 days  \n",
              "4  1.9  0.7  1.1  2022-03-06   2641691  36 years, 7 months, 25 days  "
            ]
          },
          "execution_count": 3,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "data.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "id": "1569678d",
      "metadata": {},
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x792 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x792 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x792 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x792 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x792 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# NOTE: PTS vs Salary\n",
        "plt.figure(figsize=(12,11))\n",
        "plt.xlabel('PTS', fontsize= 15)\n",
        "plt.ylabel('Salary', fontsize= 15)\n",
        "sns.scatterplot(data=data, x='PTS', y='Salary',s=65,sizes=(30,100), palette='deep')\n",
        "plt.show()\n",
        "\n",
        "\n",
        "# NOTE: REB vs Salary\n",
        "plt.figure(figsize=(12,11))\n",
        "plt.xlabel('REB', fontsize= 15)\n",
        "plt.ylabel('Salary', fontsize= 15)\n",
        "sns.scatterplot(data=data, x='REB', y='Salary',s=65,sizes=(30,100), palette='deep')\n",
        "plt.show()\n",
        "\n",
        "\n",
        "# NOTE: AST vs Salary\n",
        "plt.figure(figsize=(12,11))\n",
        "plt.xlabel('AST', fontsize= 15)\n",
        "plt.ylabel('Salary', fontsize= 15)\n",
        "sns.scatterplot(data=data, x='AST', y='Salary',s=65,sizes=(30,100), palette='deep')\n",
        "plt.show()\n",
        "\n",
        "# NOTE: STL vs Salary\n",
        "plt.figure(figsize=(12,11))\n",
        "plt.xlabel('STL', fontsize= 15)\n",
        "plt.ylabel('Salary', fontsize= 15)\n",
        "sns.scatterplot(data=data, x='STL', y='Salary',s=65,sizes=(30,100), palette='deep')\n",
        "plt.show()\n",
        "\n",
        "# NOTE: BLK vs SALARY\n",
        "plt.figure(figsize=(12,11))\n",
        "plt.xlabel('BLK', fontsize= 15)\n",
        "plt.ylabel('Salary', fontsize= 15)\n",
        "sns.scatterplot(data=data, x='BLK', y='Salary',s=65,sizes=(30,100), palette='deep')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "bbf0fe81-4aed-4dd9-8500-0919b4ec1e07",
      "metadata": {
        "id": "bbf0fe81-4aed-4dd9-8500-0919b4ec1e07"
      },
      "source": [
        "#### What can we say about these distributions?"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "36cc6b14",
      "metadata": {},
      "source": [
        "- Points vs Salary: We see a clear positive correlation between the feature (Points) and the target value (Salary). As a player, the more points you make, greater the salary you'll earn. The points has the strongest correlation against Salary. In this case, the salary explains very well the points per game for every player.\n",
        "- Reb vs Salary: There is a good correlation until the 6 rebs per game. After 6 rebs, the correlation dilutes. \n",
        "- Asist vs Salary: We don't see any correlation worth to study or observe. In the firsts mean values it seems to be a modest correlation (=> 0.5). \n",
        "- STLs / BLKs vs Salary: There is no correlation between these variables. Both are independent on the distribution against Salary."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "17735366-22bf-4cd8-a392-9dfb01652da0",
      "metadata": {
        "id": "17735366-22bf-4cd8-a392-9dfb01652da0"
      },
      "source": [
        "### Splitting the Dataset"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "02450fb0-829c-420b-b65c-ea553a9a81a2",
      "metadata": {
        "id": "02450fb0-829c-420b-b65c-ea553a9a81a2"
      },
      "source": [
        "Split the dataset into train and testing sets. "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "id": "d532e0fb-e478-48a0-971b-1ad1d428fe95",
      "metadata": {
        "id": "d532e0fb-e478-48a0-971b-1ad1d428fe95"
      },
      "outputs": [],
      "source": [
        "X = data.loc[:, data.columns != 'Salary']\n",
        "\n",
        "y = data['Salary']\n",
        "\n",
        "X_train, X_test, y_train, y_test = model_selection.train_test_split(X, y, test_size=0.20, random_state=42)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "6aec0deb-7efe-438c-8a4a-39f940aeb474",
      "metadata": {
        "id": "6aec0deb-7efe-438c-8a4a-39f940aeb474"
      },
      "source": [
        "### Baseline Model and Evaluation Metric"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "71a7eb29-2bdf-4faf-93e9-3206dbc9cca2",
      "metadata": {
        "id": "71a7eb29-2bdf-4faf-93e9-3206dbc9cca2"
      },
      "source": [
        "We need an evaluation metric and a baseline model to see how well our models fit the data. For this project we will use [Mean Absolute Error](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.mean_absolute_error.html#sklearn.metrics.mean_absolute_error) as our regression evaluation metric, because it is a really simple metric to interpret. A lower value means the model performs better. \n",
        "\n",
        "For this particular problem, we will always round the value, as we don't need decimals for this task."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "cefd619d-e6b0-4e70-acf8-f0f4a90fe896",
      "metadata": {
        "id": "cefd619d-e6b0-4e70-acf8-f0f4a90fe896"
      },
      "source": [
        "Baseline model class that implements fit() and predict() methods. As this model will not consider any other variable, the fit method should only receive a list of all players salaries. The predict method should receive and iterable and return a numpy array with the same length but every element should have the same value (the average salary)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "id": "8ee01164-1ef2-44dc-8baa-74b630444577",
      "metadata": {
        "id": "8ee01164-1ef2-44dc-8baa-74b630444577"
      },
      "outputs": [],
      "source": [
        "class BaselineModel():\n",
        "  \"\"\"\n",
        "  A baseline model that always returns the same value, the mean of the \n",
        "  players salary in the train data\n",
        "  \"\"\"\n",
        "  def __init__(self):\n",
        "    self.trained_data = 0\n",
        "  \n",
        "  def fit(self, y_train):\n",
        "    \"\"\"\n",
        "    Fit the training data. \n",
        "    \n",
        "    Parameters\n",
        "    ----------\n",
        "    y_train: Union[pd.Series, np.ndarray]\n",
        "        a pandas series or numpy array containing salary information\n",
        "    \"\"\"\n",
        "    self.trained_data = round(y_train.mean())\n",
        "\n",
        "  \n",
        "  def predict(self, X):\n",
        "    \"\"\"\n",
        "    Predict salaries \n",
        "    \n",
        "    Parameters\n",
        "    ----------\n",
        "    X: Union[pd.DataFrame, pd.Series, np.ndarray]\n",
        "        a pandas series, dataframe or numpy array with the \n",
        "        \n",
        "    Returns\n",
        "    -------\n",
        "    np.ndarray\n",
        "        a numpy array of the same length as X, with all elements equal to \n",
        "        the mean salary calculated in fit()\n",
        "    \"\"\"\n",
        "    X = np.full(len(X),self.trained_data)\n",
        "\n",
        "    return X"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "6a1a8134-48a4-4c66-84d2-39030242ef5f",
      "metadata": {
        "id": "6a1a8134-48a4-4c66-84d2-39030242ef5f"
      },
      "source": [
        "Evaluate the performance of our BaselineModel on the test set using mean_absolute_error from Sklearn (round the result)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "id": "fb72f39b-e7fc-4cd8-bd23-ee4407a80eed",
      "metadata": {
        "id": "fb72f39b-e7fc-4cd8-bd23-ee4407a80eed"
      },
      "outputs": [],
      "source": [
        "b_l_m = BaselineModel()\n",
        "\n",
        "# For this case we are using test set instead of validation set.\n",
        "\n",
        "b_l_m.fit(y_train)\n",
        "y_predicted = b_l_m.predict(X_test)\n",
        "\n",
        "baseline_mae = round(mean_absolute_error(y_test, y_predicted))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "id": "6c64f32d",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Baseline MAE: 6948493\n"
          ]
        }
      ],
      "source": [
        "print(f'Baseline MAE: {baseline_mae}')"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "fc4dff42-e36c-478d-bfd3-7541f3f36f16",
      "metadata": {
        "id": "fc4dff42-e36c-478d-bfd3-7541f3f36f16"
      },
      "source": [
        "### Regression with a single variable"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "id": "173f1211-95e2-4bcb-ba7d-a40c516b4c24",
      "metadata": {
        "id": "173f1211-95e2-4bcb-ba7d-a40c516b4c24"
      },
      "outputs": [],
      "source": [
        "scaler = StandardScaler()\n",
        "\n",
        "\n",
        "X_train[['PTS']] = scaler.fit_transform(X_train[['PTS']])\n",
        "X_test[['PTS']] = scaler.transform(X_test[['PTS']])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "id": "15d198ad-a601-4722-937b-3be52090c8d1",
      "metadata": {
        "id": "15d198ad-a601-4722-937b-3be52090c8d1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "4908976.27\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but SGDRegressor was fitted with feature names\n",
            "  warnings.warn(\n"
          ]
        }
      ],
      "source": [
        "# Create the SGDR object\n",
        "sgd_regressor = linear_model.SGDRegressor()\n",
        "\n",
        "# Fit the function to our train datasets\n",
        "\n",
        "sgd_regressor.fit(X_train[['PTS']],y_train)\n",
        "\n",
        "# Predict values\n",
        "y_predicted = sgd_regressor.predict(X_test.PTS.values.reshape(-1,1))\n",
        "\n",
        "# Calculate the metric to evaluate performance\n",
        "default_mae = round(mean_absolute_error(y_test, y_predicted),2)\n",
        "print(default_mae)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "id": "9e1f26f4-1eb4-4ecc-bfc2-9d8583f00d0d",
      "metadata": {
        "id": "9e1f26f4-1eb4-4ecc-bfc2-9d8583f00d0d"
      },
      "outputs": [],
      "source": [
        "def best_simple_regression(max_iter: int, eta0: float) -> dict:\n",
        "  \"\"\"\n",
        "  Function that tries different configurations for the hyperparameters for training: epochs, learning rate and alpha. \n",
        "\n",
        "  For each combination of parameters, evaluate the trained algorithm on the test set, then save the results of each hyperparameter combination. \n",
        "\n",
        "  Finally select the one that has the best MAE (performance metric) and print the combination and the MAE.\n",
        "\n",
        "  Parameters\n",
        "  ------------------------------\n",
        "  max_iter: int\n",
        "  eta0: float\n",
        "  \n",
        "  Return\n",
        "  ------------------------------\n",
        "  result: dict\n",
        "  \"\"\"\n",
        "  result = {\n",
        "    \"hyperparameters\": {\"max_iter\": None, \"eta0\": None}, \n",
        "    \"mae\": None\n",
        "  }\n",
        "  # Find the best hyperparameters for our model\n",
        "  mae = math.inf\n",
        "  for max_iter_1 in max_iter:\n",
        "    print(f'Trying max_iter: {max_iter_1}')\n",
        "    for eta0_1 in eta0:\n",
        "      print(f'With {max_iter_1}, trying eta0: {eta0_1}')\n",
        "      sgd_regressor = linear_model.SGDRegressor(max_iter=max_iter_1,eta0=eta0_1,random_state=3)\n",
        "      sgd_regressor.fit(X_train[['PTS']],y_train)\n",
        "      to_predict = sgd_regressor.predict(X_test[['PTS']])\n",
        "      mae_sgdr = mean_absolute_error(y_test, to_predict)\n",
        "      print(f'With max_iter: {max_iter_1} and eta0: {eta0_1}: \\nMAE: {round(mae_sgdr,3)}')\n",
        "      print(\"\")\n",
        "      if mae_sgdr < mae:\n",
        "        mae = mae_sgdr\n",
        "        result = {\n",
        "        \"hyperparameters\": {\"max_iter\": max_iter_1, \"eta0\": eta0_1}, \n",
        "        \"mae\": round(mae_sgdr,3)\n",
        "      }\n",
        "      else:\n",
        "        break\n",
        "\n",
        "  return result"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "id": "95ffbb3b-d161-4a29-9c48-eaff18b82412",
      "metadata": {
        "id": "95ffbb3b-d161-4a29-9c48-eaff18b82412"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Trying max_iter: 1000\n",
            "With 1000, trying eta0: 0.0001\n",
            "With max_iter: 1000 and eta0: 0.0001: \n",
            "MAE: 4774100.224\n",
            "\n",
            "With 1000, trying eta0: 0.02050612244897959\n",
            "With max_iter: 1000 and eta0: 0.02050612244897959: \n",
            "MAE: 4923158.74\n",
            "\n",
            "Trying max_iter: 10000\n",
            "With 10000, trying eta0: 0.0001\n",
            "With max_iter: 10000 and eta0: 0.0001: \n",
            "MAE: 4909944.012\n",
            "\n",
            "Trying max_iter: 100000\n",
            "With 100000, trying eta0: 0.0001\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\linear_model\\_stochastic_gradient.py:1527: ConvergenceWarning: Maximum number of iteration reached before convergence. Consider increasing max_iter to improve the fit.\n",
            "  warnings.warn(\n"
          ]
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "With max_iter: 100000 and eta0: 0.0001: \n",
            "MAE: 4909944.012\n",
            "\n",
            "Trying max_iter: 1000000\n",
            "With 1000000, trying eta0: 0.0001\n",
            "With max_iter: 1000000 and eta0: 0.0001: \n",
            "MAE: 4909944.012\n",
            "\n",
            "Trying max_iter: 10000000\n",
            "With 10000000, trying eta0: 0.0001\n",
            "With max_iter: 10000000 and eta0: 0.0001: \n",
            "MAE: 4909944.012\n",
            "\n",
            "Trying max_iter: 100000000\n",
            "With 100000000, trying eta0: 0.0001\n",
            "With max_iter: 100000000 and eta0: 0.0001: \n",
            "MAE: 4909944.012\n",
            "\n",
            "Best Hyperparameters for this regression {'hyperparameters': {'max_iter': 1000, 'eta0': 0.0001}, 'mae': 4774100.224}\n"
          ]
        }
      ],
      "source": [
        "# List/Array of hyperparameters to iterate over\n",
        "\n",
        "max_iter = [10**3, 10**4, 10**5, 10**6, 10**7, 10**8]\n",
        "\n",
        "eta0 = np.linspace(0.0001,1,num=50)\n",
        "\n",
        "result = best_simple_regression(max_iter, eta0)\n",
        "print('Best Hyperparameters for this regression', result)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "id": "aa909bcd",
      "metadata": {},
      "outputs": [],
      "source": [
        "# #: Regression with a single variable\n",
        "best_mae_uni = result['mae']\n",
        "best_max_iter_uni = result['hyperparameters']['max_iter']\n",
        "best_eta0_uni = result['hyperparameters']['eta0']"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "id": "776e0c7a-d592-4479-aec6-870edff61def",
      "metadata": {
        "id": "776e0c7a-d592-4479-aec6-870edff61def"
      },
      "source": [
        "Let's compare the results of the baseline, the SGDRegressor with default parameters and the fine-tuned."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "14e71ec1",
      "metadata": {},
      "source": [
        "- **As we can see, we could decrease the error (the space between predicted and truth values) by tunning the hyperparameters**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "id": "2e520ef6-52ed-48dc-8450-03031f3a03a6",
      "metadata": {
        "id": "2e520ef6-52ed-48dc-8450-03031f3a03a6"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mean Absolute Error for each model\n",
            "\n",
            "Baseline: 6948493\n",
            "Default SGDRegressor: 4908976.27\n",
            "Best SGDRegressor found with best_simple_regression(): 4774100.224\n"
          ]
        }
      ],
      "source": [
        "print(\"Mean Absolute Error for each model\")\n",
        "print(\"\")\n",
        "print(f\"Baseline: {baseline_mae}\")\n",
        "print(f\"Default SGDRegressor: {round(default_mae,2)}\")\n",
        "print(f\"Best SGDRegressor found with best_simple_regression():\", best_mae_uni)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "19e281ec-d578-496d-9f5d-897a2e1986a2",
      "metadata": {
        "id": "19e281ec-d578-496d-9f5d-897a2e1986a2"
      },
      "source": [
        "### Multivariate Linear Regression"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4ff24d36-439a-478f-b0de-0021223f6e99",
      "metadata": {
        "id": "4ff24d36-439a-478f-b0de-0021223f6e99"
      },
      "source": [
        "To improve our model, we will add more features to it. We will have to do some processing to our dataset to be able to use the data.\n",
        "Let's start with the following variables:"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "247f4716-496a-4040-9519-77f7a6839691",
      "metadata": {
        "id": "247f4716-496a-4040-9519-77f7a6839691"
      },
      "source": [
        "- Points\n",
        "- Rebounds\n",
        "- Assists\n",
        "- Blocks\n",
        "- Experience\n",
        "- Position\n",
        "- Draft Number\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "7958f7ec-45a0-4deb-bb79-3daa756dbcca",
      "metadata": {
        "id": "7958f7ec-45a0-4deb-bb79-3daa756dbcca"
      },
      "source": [
        "\n",
        "\n",
        "- [\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\", \"POSITION\", \"DRAFT_NUMBER\", \"SALARY\"]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "id": "64b265b4-d4e5-40a6-a9c3-cce920ecaa1b",
      "metadata": {
        "id": "64b265b4-d4e5-40a6-a9c3-cce920ecaa1b"
      },
      "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>PTS</th>\n",
              "      <th>REB</th>\n",
              "      <th>AST</th>\n",
              "      <th>BLK</th>\n",
              "      <th>SEASON_EXP</th>\n",
              "      <th>POSITION</th>\n",
              "      <th>DRAFT_NUMBER</th>\n",
              "      <th>Salary</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PERSON_ID</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1630173</th>\n",
              "      <td>6.5</td>\n",
              "      <td>5.1</td>\n",
              "      <td>0.8</td>\n",
              "      <td>0.5</td>\n",
              "      <td>1</td>\n",
              "      <td>Forward</td>\n",
              "      <td>20</td>\n",
              "      <td>2711280</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>203500</th>\n",
              "      <td>9.3</td>\n",
              "      <td>7.9</td>\n",
              "      <td>1.4</td>\n",
              "      <td>0.9</td>\n",
              "      <td>8</td>\n",
              "      <td>Center</td>\n",
              "      <td>12</td>\n",
              "      <td>17073171</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1628389</th>\n",
              "      <td>13.2</td>\n",
              "      <td>8.2</td>\n",
              "      <td>3.5</td>\n",
              "      <td>0.9</td>\n",
              "      <td>4</td>\n",
              "      <td>Center</td>\n",
              "      <td>14</td>\n",
              "      <td>28103550</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            PTS  REB  AST  BLK  SEASON_EXP POSITION DRAFT_NUMBER    Salary\n",
              "PERSON_ID                                                                 \n",
              "1630173     6.5  5.1  0.8  0.5           1  Forward           20   2711280\n",
              "203500      9.3  7.9  1.4  0.9           8   Center           12  17073171\n",
              "1628389    13.2  8.2  3.5  0.9           4   Center           14  28103550"
            ]
          },
          "execution_count": 41,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Create a copy of our initial DF\n",
        "mv_regression_df = data.copy()\n",
        "\n",
        "# PERSON_ID (player_id) as index\n",
        "mv_regression_df.set_index('PERSON_ID',inplace=True)\n",
        "\n",
        "# Select the following columns \n",
        "columns = [\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\", \"POSITION\", \"DRAFT_NUMBER\", \"Salary\"]\n",
        "\n",
        "# Pick our dataframe with only those we selected before.\n",
        "mv_regression_df = mv_regression_df[columns]\n",
        "mv_regression_df.head(3)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4865eb6e-d85d-4283-a67c-93df46a5b72e",
      "metadata": {
        "id": "4865eb6e-d85d-4283-a67c-93df46a5b72e"
      },
      "source": [
        "#### Feature Engineering"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "ec9b71e8-57aa-405a-8b3b-b0585a45d64d",
      "metadata": {
        "id": "ec9b71e8-57aa-405a-8b3b-b0585a45d64d"
      },
      "source": [
        "Binning the DRAFT_NUMBER feature. Draft position directly impacts salaries during the first few years, but second round players and undrafted players don't have guaranteed contracts. So we're gonna divide the bins like this: \n",
        "\n",
        "       - 1 to 15: firstround_lottery\n",
        "       - 15 to 30: firstround_non_lottery\n",
        "       - 30 to 60: second_round\n",
        "       - Undrafted: undrafted\n",
        "       "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 47,
      "id": "e117eb8e",
      "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>PTS</th>\n",
              "      <th>REB</th>\n",
              "      <th>AST</th>\n",
              "      <th>BLK</th>\n",
              "      <th>SEASON_EXP</th>\n",
              "      <th>POSITION</th>\n",
              "      <th>DRAFT_NUMBER</th>\n",
              "      <th>Salary</th>\n",
              "      <th>DRAFT</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>PERSON_ID</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>1630173</th>\n",
              "      <td>6.5</td>\n",
              "      <td>5.1</td>\n",
              "      <td>0.8</td>\n",
              "      <td>0.5</td>\n",
              "      <td>1</td>\n",
              "      <td>Forward</td>\n",
              "      <td>20</td>\n",
              "      <td>2711280</td>\n",
              "      <td>firstround_non_lottery</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>203500</th>\n",
              "      <td>9.3</td>\n",
              "      <td>7.9</td>\n",
              "      <td>1.4</td>\n",
              "      <td>0.9</td>\n",
              "      <td>8</td>\n",
              "      <td>Center</td>\n",
              "      <td>12</td>\n",
              "      <td>17073171</td>\n",
              "      <td>firstround_lottery</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1628389</th>\n",
              "      <td>13.2</td>\n",
              "      <td>8.2</td>\n",
              "      <td>3.5</td>\n",
              "      <td>0.9</td>\n",
              "      <td>4</td>\n",
              "      <td>Center</td>\n",
              "      <td>14</td>\n",
              "      <td>28103550</td>\n",
              "      <td>firstround_lottery</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            PTS  REB  AST  BLK  SEASON_EXP POSITION  DRAFT_NUMBER    Salary  \\\n",
              "PERSON_ID                                                                     \n",
              "1630173     6.5  5.1  0.8  0.5           1  Forward            20   2711280   \n",
              "203500      9.3  7.9  1.4  0.9           8   Center            12  17073171   \n",
              "1628389    13.2  8.2  3.5  0.9           4   Center            14  28103550   \n",
              "\n",
              "                            DRAFT  \n",
              "PERSON_ID                          \n",
              "1630173    firstround_non_lottery  \n",
              "203500         firstround_lottery  \n",
              "1628389        firstround_lottery  "
            ]
          },
          "execution_count": 47,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Replace Undrafted with 0\n",
        "mv_regression_df['DRAFT_NUMBER'].replace('Undrafted',0,inplace=True)\n",
        "\n",
        "# Fill NaN with 0\n",
        "mv_regression_df.fillna(0,inplace=True)\n",
        "\n",
        "# Cast to integer\n",
        "mv_regression_df.loc[:,'DRAFT_NUMBER'] = mv_regression_df['DRAFT_NUMBER'].astype(int)\n",
        "\n",
        "# Bins\n",
        "bins = [0,1,15,30,60]\n",
        "\n",
        "# List of labels \n",
        "labels = ['undrafted','firstround_lottery','firstround_non_lottery','second_round']\n",
        "\n",
        "\n",
        "mv_regression_df['DRAFT'] = pd.cut(mv_regression_df['DRAFT_NUMBER'], bins=bins, include_lowest=True ,labels=labels)\n",
        "mv_regression_df.head(3)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "37a55237-cb04-4c38-bef3-8c363f9dbe5c",
      "metadata": {
        "id": "37a55237-cb04-4c38-bef3-8c363f9dbe5c"
      },
      "source": [
        "Encode the categorical features: DRAFT, POSITION"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 49,
      "id": "c565394e-5d44-4418-a33b-eaa801062dbc",
      "metadata": {
        "id": "c565394e-5d44-4418-a33b-eaa801062dbc"
      },
      "outputs": [],
      "source": [
        "mv_regression_df = pd.get_dummies(data=mv_regression_df,columns=['DRAFT','POSITION'], drop_first=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "9816d9d4-b44d-4bb7-9cad-156daee9dc50",
      "metadata": {
        "id": "9816d9d4-b44d-4bb7-9cad-156daee9dc50"
      },
      "source": [
        "#### Modeling"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "1eb528f1-852e-428e-b97c-40b412985f98",
      "metadata": {
        "id": "1eb528f1-852e-428e-b97c-40b412985f98"
      },
      "source": [
        "Split the dataset into train and test. "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 59,
      "id": "7696efd0-94f8-49ce-846f-4fd24a01b0fd",
      "metadata": {
        "id": "7696efd0-94f8-49ce-846f-4fd24a01b0fd"
      },
      "outputs": [],
      "source": [
        "X = mv_regression_df.loc[:, (mv_regression_df.columns != 'Salary') & (mv_regression_df.columns != 'DRAFT_NUMBER')]\n",
        "\n",
        "y = mv_regression_df['Salary']\n",
        "\n",
        "X_train, X_test, y_train, y_test = model_selection.train_test_split(X, y, test_size=0.20, random_state=42)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "b2540df8-ac16-4ad1-b8cf-028f566faa06",
      "metadata": {
        "id": "b2540df8-ac16-4ad1-b8cf-028f566faa06"
      },
      "source": [
        "Feature scaling in all the numerical features (\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\") except for draft number."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 61,
      "id": "94c1b356-8f74-4a20-8b9d-9dd61b6e3dec",
      "metadata": {
        "id": "94c1b356-8f74-4a20-8b9d-9dd61b6e3dec"
      },
      "outputs": [],
      "source": [
        "scaler = StandardScaler()\n",
        "\n",
        "# Scale data (fit & transform) in X_train dataset\n",
        "\n",
        "X_train[[\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\"]] = scaler.fit_transform(X_train[[\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\"]])\n",
        "\n",
        "# Idem with X_test\n",
        "\n",
        "X_test[[\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\"]] = scaler.transform(X_test[[\"PTS\", \"REB\", \"AST\", \"BLK\", \"SEASON_EXP\"]])\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 62,
      "id": "7d8642f3-895c-4d8d-85b7-80a087a0adef",
      "metadata": {
        "id": "7d8642f3-895c-4d8d-85b7-80a087a0adef"
      },
      "outputs": [],
      "source": [
        "def best_mv_regression(max_iter, eta0):\n",
        "  result_mv = {\n",
        "    \"hyperparameters\": {\"max_iter\": None, \"eta0\": None}, \n",
        "    \"mae\": None\n",
        "  }\n",
        "  mae = math.inf\n",
        "  for max_iter_1 in max_iter:\n",
        "    print(f'Trying max_iter: {max_iter_1}')\n",
        "    for eta0_1 in eta0:\n",
        "      print(f'With {max_iter_1}, trying eta0: {eta0_1}')\n",
        "      sgd_regressor = linear_model.SGDRegressor(max_iter=max_iter_1,eta0=eta0_1,random_state=7)\n",
        "      sgd_regressor.fit(X_train,y_train)\n",
        "      y_predicted = sgd_regressor.predict(X_test)\n",
        "      mae_sgdr = mean_absolute_error(y_test, y_predicted)\n",
        "      print(f'With max_iter: {max_iter_1} and eta0: {eta0_1}: \\nMAE: {round(mae_sgdr,3)}')\n",
        "      print(\"\")\n",
        "      if mae_sgdr < mae:\n",
        "        mae = mae_sgdr\n",
        "        result_mv = {\n",
        "        \"hyperparameters\": {\"max_iter\": max_iter_1, \"eta0\": eta0_1}, \n",
        "        \"mae\": round(mae_sgdr,2)\n",
        "      }\n",
        "\n",
        "  return result_mv"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 63,
      "id": "a2688875",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Set space/values for hyperparameters to iterate\n",
        "\n",
        "max_iter = [10**3, 10**4, 10**5, 10**6, 10**7, 10**8]\n",
        "\n",
        "eta0 = np.linspace(0.0001,1,num=50)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "6ca72f97",
      "metadata": {},
      "outputs": [],
      "source": [
        "result_mv = best_mv_regression(max_iter, eta0)\n",
        "print(result_mv)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 65,
      "id": "2108de73",
      "metadata": {},
      "outputs": [],
      "source": [
        "best_mae_mv = result_mv['mae']\n",
        "best_max_iter_mv = result_mv['hyperparameters']['max_iter']\n",
        "best_eta0_mv = result_mv['hyperparameters']['eta0']"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "33d63bc7-448d-40a6-a314-6602c75e2284",
      "metadata": {
        "id": "33d63bc7-448d-40a6-a314-6602c75e2284"
      },
      "source": [
        "How did the model performed compared to our univariate models? "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 66,
      "id": "169eb203",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Univariate model performed better against multivariate, with a MAE of 4774100.224\n",
            "The Multivariate MAE (4945258.49) is greater by 171158.27\n"
          ]
        }
      ],
      "source": [
        "print(f'Univariate model performed better against multivariate, with a MAE of {best_mae_uni}')\n",
        "print(f'The Multivariate MAE ({best_mae_mv}) is greater by {round((best_mae_mv - best_mae_uni),2)}')"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "id": "e9992197-563b-4a69-9cd7-97fcb9ee3677",
      "metadata": {
        "id": "e9992197-563b-4a69-9cd7-97fcb9ee3677",
        "tags": []
      },
      "source": [
        "### DecisionTreeRegressor\n",
        "\n",
        "It is possible this problem might not be easily resolvable by a linear model, or that there could be better algorithms to tackle it. As an example, I'll grab a very used algorithm, Decision Tree."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 67,
      "id": "f54443e3",
      "metadata": {},
      "outputs": [],
      "source": [
        "dt_regression_df = mv_regression_df.copy()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 68,
      "id": "023c5a29",
      "metadata": {},
      "outputs": [],
      "source": [
        "# #: Select features & labels and then split train/test datasets.\n",
        "\n",
        "X = dt_regression_df.loc[:, (dt_regression_df.columns != 'Salary') & (dt_regression_df.columns != 'DRAFT_NUMBER')]\n",
        "\n",
        "y = dt_regression_df['Salary']\n",
        "\n",
        "X_train, X_test, y_train, y_test = model_selection.train_test_split(X, y, test_size=0.20, random_state=35)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 70,
      "id": "c69ac038",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Instantiate DecisionTree with default hyperparameters\n",
        "d_tree = DecisionTreeRegressor()\n",
        "\n",
        "# Fit our model\n",
        "d_tree.fit(X_train,y_train)\n",
        "\n",
        "# Predict values\n",
        "y_predicted = d_tree.predict(X_test)\n",
        "\n",
        "d_tree_accuracy_score_model = d_tree.score(X_test,y_predicted)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 76,
      "id": "182baad5",
      "metadata": {},
      "outputs": [],
      "source": [
        "def best_dt_regressor(max_depth: int, min_samples_leaf: int , max_features: int) -> dict:\n",
        "  result = {\n",
        "    \"hyperparameters\": {\"max_depth\": None, \"min_samples_leaf\": None, \"max_features\": None}, \n",
        "    \"mae\": None\n",
        "  }\n",
        "  mae = math.inf\n",
        "  for max_features_iter in max_features:\n",
        "      print(f'Trying max_features: {max_features_iter}')\n",
        "      for max_depth_iter in max_depth:\n",
        "        print(f'With max_features: {max_features_iter}, trying max_depth: {max_depth_iter}')\n",
        "        for min_samples_leaf_iter in min_samples_leaf:\n",
        "          print(f'With max_features {max_features_iter} and max_depth: {max_depth_iter}, trying min_samples_leaf: {min_samples_leaf_iter}')\n",
        "          d_tree = DecisionTreeRegressor(max_features=max_features_iter,max_depth=max_depth_iter,min_samples_leaf=min_samples_leaf_iter)\n",
        "          d_tree.fit(X_train,y_train)\n",
        "          y_predicted = d_tree.predict(X_test)\n",
        "          mae_sgdr = mean_absolute_error(y_test, y_predicted)\n",
        "          print(f'With max_features {max_features_iter}, max_depth: {max_depth_iter}, trying min_samples_leaf: {min_samples_leaf_iter} \\n MAE: {round(mae_sgdr,3)}')\n",
        "          print(\"\")\n",
        "          if mae_sgdr < mae:\n",
        "            mae = mae_sgdr\n",
        "            result = {\n",
        "            \"hyperparameters\": {\"max_depth\": max_depth_iter, \"min_samples_leaf\": min_samples_leaf_iter, \"max_features\": max_features_iter}, \n",
        "            \"Best MAE\": round(mae,2)\n",
        "          }\n",
        "\n",
        "  return result"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "1776c34b",
      "metadata": {},
      "outputs": [],
      "source": [
        "# List of hyperparameter example values\n",
        "max_depth = [5, 10, 15, 20, 50] # Quantity of \"generations\"?\n",
        "min_samples_leaf = [2, 10, 20, 50] # Quantity of samples per \"children\"?\n",
        "max_features = [5, 10]\n",
        "\n",
        "result_dtree = best_dt_regressor(max_depth, min_samples_leaf,max_features)\n",
        "print(result_dtree)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 79,
      "id": "c57389e2",
      "metadata": {},
      "outputs": [],
      "source": [
        "best_mae_dt = result_dtree['Best MAE']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 80,
      "id": "af1084d8-97a4-4618-b4f4-0952e0b39f09",
      "metadata": {
        "id": "af1084d8-97a4-4618-b4f4-0952e0b39f09"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mean Absolute Error for each model\n",
            "\n",
            "Baseline: 6948493\n",
            "Default SGDRegressor: 4908976.27\n",
            "Best SGDRegressor found with best_simple_regression(): 4774100.224\n",
            "Best DTRegressor found with best_dt_regressor(): 3780848.65\n"
          ]
        }
      ],
      "source": [
        "print(\"Mean Absolute Error for each model\")\n",
        "print(\"\")\n",
        "print(f\"Baseline: {baseline_mae}\")\n",
        "print(f\"Default SGDRegressor: {round(default_mae,2)}\")\n",
        "print(f\"Best SGDRegressor found with best_simple_regression():\", best_mae_uni)\n",
        "print(f\"Best DTRegressor found with best_dt_regressor():\", best_mae_dt)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "ce424a87-37af-470d-9a14-ec6ae2258eb1",
      "metadata": {
        "id": "ce424a87-37af-470d-9a14-ec6ae2258eb1"
      },
      "source": [
        "### Analyzing our results"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e204456d-cb86-47a0-9c3c-6307d8a831fa",
      "metadata": {
        "id": "e204456d-cb86-47a0-9c3c-6307d8a831fa"
      },
      "source": [
        "Which kind of model worked better in this dataset, linear or non linear models? Do you have any hypothesis about why? "
      ]
    },
    {
      "cell_type": "markdown",
      "id": "5188dc61",
      "metadata": {},
      "source": [
        "- **Decision Tree** outperforms linear models. The non linear model work really good with this data layout. DTrees understand the behaviour between non-linear relations, make a comparisson between variables and pick the best outcome.     "
      ]
    },
    {
      "cell_type": "markdown",
      "id": "95ea5922",
      "metadata": {},
      "source": [
        "##### Tasks to improve model accuracy:\n",
        "1. Add more data: It's always a good idea to deliver more data in order to let our model learn more (be aware of overfitting).\n",
        "2. Special treatment to missing values and outliers: It's important to keep away any bias to our model. Missing values and outliers increase this risk.\n",
        "**A** Missing values: Replace those with median, mode or mean. \n",
        "**B** Outliers: Eliminate from the model. Try binning the data.\n",
        "3. Feature engineering: Our model (and humans also) extract information from data. Invest time and quality tasks to clean, sort, bin and normalize data. \n",
        "4. Scale data: Try differents scale algorithms and leverage the scaling.\n",
        "5. Feature selection: Observe and pick wisely the features. Plot data, gain insights and understand better what's happening about your variable distribution. \n",
        "\n",
        "\n",
        "    "
      ]
    },
    {
      "cell_type": "markdown",
      "id": "25e23c2a-a0bf-4d39-bc1b-9524107f3a02",
      "metadata": {
        "id": "25e23c2a-a0bf-4d39-bc1b-9524107f3a02",
        "tags": []
      },
      "source": [
        "### 2. Star or no Star? Predicting players All-NBA selections"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "id": "26cd4fec-1b8b-4bfc-8024-dcdd5b8b803a",
      "metadata": {
        "id": "26cd4fec-1b8b-4bfc-8024-dcdd5b8b803a"
      },
      "source": [
        "- Being selected as part of the [All-NBA teams](https://en.wikipedia.org/wiki/All-NBA_Team) is different than being selected for [All-Star Game](https://en.wikipedia.org/wiki/NBA_All-Star_Game). \n",
        "- Here, we're going to predict All-NBA selections, which takes place after the regular season but before playoffs start, so we're only going to use players and teams statistics about the regular season. \n",
        "- All Star game takes place at the middle point of the regular season, players are selected based on partial season performance, and the general public has 50% of the votes, so it is also a more popularity-based award.\n",
        "- All-NBA teams are divided in first, second and third teams of 5 players each, divided by position: 2 front court players (guards), 2 backcourt players (forwards) and 1 center.\n",
        "- As only 15 players are selected to All-NBA teams each year (out of more than 400 qualifying players), our dataset is going to be imbalanced. We're going to have to find a good strategy to deal with this inbalance context."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "ef2353fa-10d7-48d1-85d7-f8743ee86698",
      "metadata": {
        "id": "ef2353fa-10d7-48d1-85d7-f8743ee86698"
      },
      "source": [
        "#### Analyzing the data"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "d4159ce5-fec3-48bb-93c5-058f2cbb55c5",
      "metadata": {
        "id": "d4159ce5-fec3-48bb-93c5-058f2cbb55c5"
      },
      "source": [
        "To do this task, we're going to need historical information about All-NBA selections. We'll use a dataset that contains information about players from season starting in 1990 and ending in the 2018-2019 season.\n",
        "The dataset contains statistics for all players, and a column named \"all-nba\" that is going to be our target variable, 1 means the player was selected to an All-NBA team that season."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "id": "bb6725f6-418d-4159-b046-3bf33a8ead0b",
      "metadata": {
        "id": "bb6725f6-418d-4159-b046-3bf33a8ead0b"
      },
      "outputs": [],
      "source": [
        "all_nba_df = pd.read_csv(\"all_nba_1990_2017_dataset.csv\", index_col=0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 128,
      "id": "b9b595da",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Total number of rows: 14825\n",
            "Total number of seasons: 39\n",
            "Total number of unique players: 2718\n",
            "Total NBA Players selected: 496 vs total NBA Players not selected: 14329\n"
          ]
        }
      ],
      "source": [
        "all_nba_selected = len(all_nba_df[all_nba_df['all_nba'] == 1])\n",
        "all_nba_not_selected = len(all_nba_df[all_nba_df['all_nba'] == 0])\n",
        "\n",
        "print(\"Total number of rows:\", len(all_nba_df.values))\n",
        "print(\"Total number of seasons:\", len(all_nba_df['season'].unique()))\n",
        "print(\"Total number of unique players:\", len(all_nba_df['player_id'].unique()))\n",
        "print(f\"Total NBA Players selected: {all_nba_selected} vs total NBA Players not selected: {all_nba_not_selected}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 83,
      "id": "24575b27",
      "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>season_id</th>\n",
              "      <th>player_id</th>\n",
              "      <th>player_season</th>\n",
              "      <th>player</th>\n",
              "      <th>season</th>\n",
              "      <th>season_start</th>\n",
              "      <th>team</th>\n",
              "      <th>games</th>\n",
              "      <th>gamesstarted</th>\n",
              "      <th>mp</th>\n",
              "      <th>...</th>\n",
              "      <th>2p_perc</th>\n",
              "      <th>ft_perc</th>\n",
              "      <th>trb</th>\n",
              "      <th>ast</th>\n",
              "      <th>stl</th>\n",
              "      <th>blk</th>\n",
              "      <th>tov</th>\n",
              "      <th>pf</th>\n",
              "      <th>pts</th>\n",
              "      <th>all_nba</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>Alaa Abdelnaby</td>\n",
              "      <td>1990-91</td>\n",
              "      <td>1990</td>\n",
              "      <td>POR</td>\n",
              "      <td>43</td>\n",
              "      <td>0</td>\n",
              "      <td>6.7</td>\n",
              "      <td>...</td>\n",
              "      <td>0.474</td>\n",
              "      <td>0.568</td>\n",
              "      <td>2.1</td>\n",
              "      <td>0.3</td>\n",
              "      <td>0.1</td>\n",
              "      <td>0.3</td>\n",
              "      <td>0.5</td>\n",
              "      <td>0.9</td>\n",
              "      <td>3.1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>Alaa Abdelnaby</td>\n",
              "      <td>1991-92</td>\n",
              "      <td>1991</td>\n",
              "      <td>POR</td>\n",
              "      <td>71</td>\n",
              "      <td>1</td>\n",
              "      <td>13.2</td>\n",
              "      <td>...</td>\n",
              "      <td>0.493</td>\n",
              "      <td>0.752</td>\n",
              "      <td>3.7</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.2</td>\n",
              "      <td>0.9</td>\n",
              "      <td>1.9</td>\n",
              "      <td>6.1</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>3</td>\n",
              "      <td>Alaa Abdelnaby</td>\n",
              "      <td>1992-93</td>\n",
              "      <td>1992</td>\n",
              "      <td>TOT</td>\n",
              "      <td>75</td>\n",
              "      <td>52</td>\n",
              "      <td>17.5</td>\n",
              "      <td>...</td>\n",
              "      <td>0.519</td>\n",
              "      <td>0.759</td>\n",
              "      <td>4.5</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.3</td>\n",
              "      <td>0.3</td>\n",
              "      <td>1.3</td>\n",
              "      <td>2.5</td>\n",
              "      <td>7.7</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>3 rows × 22 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "   season_id  player_id  player_season          player   season  season_start  \\\n",
              "0          0          0              1  Alaa Abdelnaby  1990-91          1990   \n",
              "1          1          0              2  Alaa Abdelnaby  1991-92          1991   \n",
              "2          2          0              3  Alaa Abdelnaby  1992-93          1992   \n",
              "\n",
              "  team  games  gamesstarted    mp  ...  2p_perc  ft_perc  trb  ast  stl  blk  \\\n",
              "0  POR     43             0   6.7  ...    0.474    0.568  2.1  0.3  0.1  0.3   \n",
              "1  POR     71             1  13.2  ...    0.493    0.752  3.7  0.4  0.4  0.2   \n",
              "2  TOT     75            52  17.5  ...    0.519    0.759  4.5  0.4  0.3  0.3   \n",
              "\n",
              "   tov   pf  pts  all_nba  \n",
              "0  0.5  0.9  3.1        0  \n",
              "1  0.9  1.9  6.1        0  \n",
              "2  1.3  2.5  7.7        0  \n",
              "\n",
              "[3 rows x 22 columns]"
            ]
          },
          "execution_count": 83,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "all_nba_df.head(3)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "f268db59-8c9e-446b-be19-290e608169d2",
      "metadata": {
        "id": "f268db59-8c9e-446b-be19-290e608169d2"
      },
      "source": [
        "### Splitting the dataset"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 85,
      "id": "25cfc4ee-98e7-4b84-98d3-80a91fb44912",
      "metadata": {
        "id": "25cfc4ee-98e7-4b84-98d3-80a91fb44912"
      },
      "outputs": [],
      "source": [
        "columns = ['team', 'games', 'gamesstarted', 'mp', 'fg_perc', '3p_perc','2p_perc', 'ft_perc', 'trb', 'ast', 'stl', 'blk', 'tov', 'pf', 'pts']\n",
        "\n",
        "X = all_nba_df[columns]\n",
        "\n",
        "y = all_nba_df['all_nba']\n",
        "\n",
        "X_train, X_test, y_train, y_test = model_selection.train_test_split(X, y, test_size=0.30, random_state=17)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "7043db6e-da74-4a04-91d9-4dd0cd830f22",
      "metadata": {
        "id": "7043db6e-da74-4a04-91d9-4dd0cd830f22"
      },
      "source": [
        "### Baseline model and Evaluation Metrics"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 132,
      "id": "350ed4e7",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Accuracy if the model always predict 0 (not being selected by all nba): \n",
            "96.7%\n"
          ]
        }
      ],
      "source": [
        "base_line_accuracy_allnba = 1 - np.sum(all_nba_df.all_nba)/all_nba_df.shape[0]\n",
        "print(f'Accuracy if the model always predict 0 (not being selected by all nba): \\n{round(base_line_accuracy_allnba,3)*100}%')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 95,
      "id": "f9f85bfc",
      "metadata": {},
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 504x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "f, ax = plt.subplots(figsize=(7, 5))\n",
        "sns.countplot(x='all_nba', data=all_nba_df)\n",
        "_ = plt.title('All NBA Selected')\n",
        "_ = plt.xlabel('Selection')"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "7b86a780",
      "metadata": {},
      "source": [
        "- Our dataset es very imbalanced, about 96.7 % of our data refers to the non-selected players."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 96,
      "id": "e4d2a505-f193-4b94-89c3-56cba2c4db54",
      "metadata": {
        "id": "e4d2a505-f193-4b94-89c3-56cba2c4db54"
      },
      "outputs": [],
      "source": [
        "scaler = StandardScaler()\n",
        "\n",
        "X_train.loc[:, (X_train.columns != 'team')] = scaler.fit_transform(X_train.loc[:, (X_train.columns != 'team')])\n",
        "\n",
        "\n",
        "X_test.loc[:, (X_test.columns != 'team')] = scaler.transform(X_test.loc[:, (X_test.columns != 'team')])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 97,
      "id": "6f9ffef5-6333-481e-816f-9125c95fe117",
      "metadata": {
        "id": "6f9ffef5-6333-481e-816f-9125c95fe117"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "<style>#sk-container-id-1 {color: black;background-color: white;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>LogisticRegression()</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" checked><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">LogisticRegression</label><div class=\"sk-toggleable__content\"><pre>LogisticRegression()</pre></div></div></div></div></div>"
            ],
            "text/plain": [
              "LogisticRegression()"
            ]
          },
          "execution_count": 97,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "logistic_regressor = LogisticRegression()\n",
        "\n",
        "logistic_regressor.fit(X_train.loc[:, (X_train.columns != 'team')],y_train)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "0f06758a-f2dd-4339-b170-275b428fdb4e",
      "metadata": {
        "id": "0f06758a-f2dd-4339-b170-275b428fdb4e"
      },
      "source": [
        "- Let's Evaluate our baseline model using this metrics: \n",
        "\n",
        "    - accuracy\n",
        "    - precision\n",
        "    - recall\n",
        "    - f1"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 133,
      "id": "b953324f-f7ae-4823-a801-f06d27d6eb43",
      "metadata": {
        "id": "b953324f-f7ae-4823-a801-f06d27d6eb43"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Evaluation of the baseline classifier:\n",
            "\n",
            "Accuracy: 0.984\n",
            "Precision: 0.788\n",
            "Recall: 0.669\n",
            "F1: 0.724\n"
          ]
        }
      ],
      "source": [
        "# #: Excluiding 'team' from datasets\n",
        "print(\"\")\n",
        "print('Evaluation of the baseline classifier:')\n",
        "print(\"\")\n",
        "\n",
        "y_predicted = logistic_regressor.predict(X_test.loc[:, (X_test.columns != 'team')])\n",
        "\n",
        "# #: Accuracy\n",
        "\n",
        "blm_accuracy = accuracy_score(y_test,y_predicted)\n",
        "print(f'Accuracy: {round(blm_accuracy,3)}')\n",
        "\n",
        "# #: Precision\n",
        "\n",
        "blm_precision = precision_score(y_test,y_predicted)\n",
        "print(f'Precision: {round(blm_precision,3)}')\n",
        "\n",
        "# #: Recall -> The ability of the classifier to find all the POSITIVE samples. In other words, how well detects the True Positives.\n",
        "\n",
        "blm_recall = recall_score(y_test,y_predicted)\n",
        "print(f'Recall: {round(blm_recall,3)}')\n",
        "\n",
        "# #: F1\n",
        "\n",
        "blm_f1 = f1_score(y_test,y_predicted)\n",
        "print(f'F1: {round(blm_f1,3)}')\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "2054a43f-55ec-45fc-9ead-a7699d8aa2b0",
      "metadata": {
        "id": "2054a43f-55ec-45fc-9ead-a7699d8aa2b0"
      },
      "source": [
        "### Classification"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "1e2847f3-f02c-4308-8a0c-41ba79af0781",
      "metadata": {
        "id": "1e2847f3-f02c-4308-8a0c-41ba79af0781"
      },
      "source": [
        "#### Feature engineering"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "27b6ddb3-576c-4238-9ec3-534aafee764b",
      "metadata": {
        "id": "27b6ddb3-576c-4238-9ec3-534aafee764b"
      },
      "source": [
        "One hot encode the Team feature"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 134,
      "id": "f4ca83ec-7aab-4cff-8ac7-1b224ff54dde",
      "metadata": {
        "id": "f4ca83ec-7aab-4cff-8ac7-1b224ff54dde"
      },
      "outputs": [],
      "source": [
        "# Encode X_train in team column\n",
        "\n",
        "X_train = pd.get_dummies(data=X_train, columns=['team'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 135,
      "id": "d8484c88",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Encode X_test in team column\n",
        "\n",
        "X_test = pd.get_dummies(data=X_test,columns=['team'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 136,
      "id": "e7a96b18",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Drop team_TOT \n",
        "\n",
        "X_train.drop(columns=['team_TOT'], inplace=True)\n",
        "X_test.drop(columns=['team_TOT'], inplace=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "6592c6c0-7ec8-49ac-936c-2d0fb84ccfef",
      "metadata": {
        "id": "6592c6c0-7ec8-49ac-936c-2d0fb84ccfef",
        "tags": []
      },
      "source": [
        "#### Modeling"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 137,
      "id": "4bdc4abf-ded3-4d8b-9170-c1d5b4951ba7",
      "metadata": {
        "id": "4bdc4abf-ded3-4d8b-9170-c1d5b4951ba7"
      },
      "outputs": [],
      "source": [
        "def search_best_hypers_logistic(tol: float, C: float, max_iter: int) -> dict:\n",
        "  \"\"\"\n",
        "  Train a logistic regresion model, find the best hyperparameters for: tols, C and max_iter. \n",
        "  \n",
        "  F1 score based as eval metric, but report also precision and recall. \n",
        "  \n",
        "  Save the best performing model to use.\n",
        "\n",
        "  Parameters\n",
        "  ---------------------\n",
        "  tol: float\n",
        "  C: float\n",
        "  max_iter: int\n",
        "\n",
        "  Returns\n",
        "  ---------------------\n",
        "  result_lr: dict\n",
        "\n",
        "  \"\"\"\n",
        "  result_lr = {\n",
        "    \"hyperparameters\": {\"tol\": None, \"C\": None, \"max_iter\": None}, \n",
        "    \"f1_score\": None\n",
        "  }\n",
        "  best_f1_score = 0\n",
        "  for tol_iter in tol:\n",
        "\n",
        "    print(f'Trying tol: {tol_iter}')\n",
        "\n",
        "    for C_iter in C:\n",
        "\n",
        "      print(f'With {tol_iter}, trying C: {C_iter}')\n",
        "\n",
        "      for max_iter_1 in max_iter:\n",
        "\n",
        "        print(f'With {tol_iter} and C: {C_iter}, trying {max_iter_1}')\n",
        "\n",
        "        logistic_regressor = LogisticRegression(tol=tol_iter, C= C_iter, max_iter=max_iter_1)\n",
        "        logistic_regressor.fit(X_train,y_train)\n",
        "        y_predicted = logistic_regressor.predict(X_test)\n",
        "\n",
        "        f1 = f1_score(y_test, y_predicted)\n",
        "        best_lr_precision = precision_score(y_test,y_predicted)\n",
        "        print(f'Precision: {round(best_lr_precision,3)}')\n",
        "\n",
        "        best_lr_recall = recall_score(y_test,y_predicted)\n",
        "        print(f'Recall: {round(best_lr_recall,3)}')\n",
        "\n",
        "        print(f'With {tol_iter} and C: {C_iter}, trying {max_iter_1} \\n  F1 LEARNED: {f1}')\n",
        "        if best_f1_score < f1:\n",
        "          best_f1_score = f1\n",
        "          result_lr = {\n",
        "          \"hyperparameters\": {\"tol\": tol_iter, \"C\": C_iter, \"max_iter_1\": max_iter_1}, \n",
        "          \"f1_score\": round(f1,2), \"Precision Score\" : round(best_lr_precision,3), \"Recall\": round(best_lr_recall,3)\n",
        "        }\n",
        "\n",
        "  return result_lr"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "c3b89ea7",
      "metadata": {},
      "outputs": [],
      "source": [
        "tol = np.linspace(0.0001,1,num=50)\n",
        "C = [0.001, 0.01, 0.1, 1, 10, 100, 1000]\n",
        "max_iter = [10**3, 10**4, 10**5, 10**6, 10**7, 10**8]\n",
        "\n",
        "\n",
        "result_lr = search_best_hypers_logistic(tol, C, max_iter)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 139,
      "id": "8e1a7aa1",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{'hyperparameters': {'tol': 0.0001, 'C': 1, 'max_iter_1': 1000}, 'f1_score': 0.75, 'Precision Score': 0.808, 'Recall': 0.698}\n"
          ]
        }
      ],
      "source": [
        "print(result_lr)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "45c0c3bd-b431-40f1-abbb-4e90b7479411",
      "metadata": {
        "id": "45c0c3bd-b431-40f1-abbb-4e90b7479411"
      },
      "source": [
        "#### Evaluating on New Data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 140,
      "id": "da0dcf05",
      "metadata": {},
      "outputs": [],
      "source": [
        "best_tol_lr= result_lr['hyperparameters']['tol']\n",
        "best_C_lr = result_lr['hyperparameters']['C']\n",
        "best_max_iter_lr = result_lr['hyperparameters']['max_iter_1']\n",
        "best_f1_lr = result_lr['f1_score']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "id": "79ea153a-b061-4df9-9c3d-1aaa0526c700",
      "metadata": {
        "id": "79ea153a-b061-4df9-9c3d-1aaa0526c700"
      },
      "outputs": [],
      "source": [
        "all_nba_2018_df = pd.read_csv('all_nba_2018_dataset.csv', index_col=0)\n",
        "all_nba_2018_selections = pd.read_csv('all_nba_2018_selections.csv', index_col=0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 142,
      "id": "14b936f3",
      "metadata": {},
      "outputs": [],
      "source": [
        "# #: Define columns to not use\n",
        "columns_unused = ['season_id', 'player_id', 'player_season', 'player', 'season', 'season_start','team']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 143,
      "id": "8dc93828",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Scale just numerical features\n",
        "\n",
        "all_nba_2018_df.loc[:, ~all_nba_2018_df.columns.isin(columns_unused)] = scaler.transform(all_nba_2018_df.loc[:, ~all_nba_2018_df.columns.isin(columns_unused)])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 144,
      "id": "6d0d2962",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Pick 40 teams\n",
        "teams_full = pd.DataFrame(all_nba_df.team.unique())\n",
        "\n",
        "# Pick 30 teams\n",
        "teams_2018 = pd.DataFrame(all_nba_2018_df['team'].unique()) \n",
        "\n",
        "# Merge to obtain full 40 teams \n",
        "intersection = pd.merge(teams_full, teams_2018, how='outer', indicator=True) \n",
        "\n",
        "# Pick the 10 extra teams\n",
        "antijoin = intersection[~(intersection['_merge']=='both')].drop('_merge', axis=1) \n",
        "\n",
        "# Create a df with those extra teams\n",
        "ten_teams_new = pd.DataFrame(antijoin,  columns=antijoin) \n",
        "\n",
        "# Fill empty rows with 0\n",
        "ten_teams_new.fillna(0,inplace=True)\n",
        "\n",
        "# Convert rows (names of teams) in columns\n",
        "for column in ten_teams_new.columns: \n",
        "    all_nba_2018_df[column] = ten_teams_new[column] "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 145,
      "id": "0471ba31",
      "metadata": {},
      "outputs": [],
      "source": [
        "# Encode 40 teams, drop rows with Nan values and columns we don't need.\n",
        "\n",
        "# Encode teams\n",
        "all_nba_2018_df = pd.get_dummies(data=all_nba_2018_df, columns=['team']) \n",
        "\n",
        "# Drop team_TOT to best fit the model\n",
        "all_nba_2018_df.drop(columns=['team_TOT'], inplace=True) \n",
        "\n",
        "# fill with 0 empty rows\n",
        "all_nba_2018_df.fillna(0,inplace=True) \n",
        "\n",
        "# pick just what you need\n",
        "all_nba_2018_df = all_nba_2018_df.loc[:, ~all_nba_2018_df.columns.isin(columns_unused)] "
      ]
    },
    {
      "cell_type": "markdown",
      "id": "8b0bb296-b1b3-4ae4-b199-b11224c78a82",
      "metadata": {
        "id": "8b0bb296-b1b3-4ae4-b199-b11224c78a82"
      },
      "source": [
        "Predict and evaluate the model using precision, recall and f1 score. "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 147,
      "id": "b00ddffa-3f09-4294-beb5-098c2d698af0",
      "metadata": {
        "id": "b00ddffa-3f09-4294-beb5-098c2d698af0"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\utils\\validation.py:1858: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['str', 'tuple']. An error will be raised in 1.2.\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names\n",
            "  warnings.warn(\n"
          ]
        }
      ],
      "source": [
        "# Predict with the best regressor\n",
        "\n",
        "best_lr = LogisticRegression(tol= 0.0001, C= 1, max_iter= 1000)\n",
        "best_lr.fit(X_train,y_train)\n",
        "y_predicted = best_lr.predict(all_nba_2018_df)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 148,
      "id": "188d55e6",
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Precision: 0.625\n",
            "\n",
            "Recall: 1.0\n",
            "\n",
            "F1: 0.769\n",
            "\n"
          ]
        }
      ],
      "source": [
        "# #: Precision\n",
        "print(\"\")\n",
        "precision_new = precision_score(all_nba_2018_selections.all_nba,y_predicted)\n",
        "print(f'Precision: {round(precision_new,3)}')\n",
        "print(\"\")\n",
        "\n",
        "# #: Recall -> The ability of the classifier to find all the POSITIVE samples. In other words, how well detects the True Positives.\n",
        "\n",
        "recall_new = recall_score(all_nba_2018_selections.all_nba,y_predicted)\n",
        "print(f'Recall: {round(recall_new,3)}')\n",
        "print(\"\")\n",
        "\n",
        "# #: F1\n",
        "\n",
        "f1_new = f1_score(all_nba_2018_selections.all_nba,y_predicted)\n",
        "print(f'F1: {round(f1_new,3)}')\n",
        "print(\"\")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "17f87aa8-dd84-462a-b26b-80400e4a87e3",
      "metadata": {
        "id": "17f87aa8-dd84-462a-b26b-80400e4a87e3"
      },
      "source": [
        "- Our dataset is imbalanced and the positive class (selected by all_nba) is the minority class. There are a lot of negative examples that could be converted into False Positives."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "77a8c607-5df7-43d2-8365-cbbb6f2b93e1",
      "metadata": {
        "id": "77a8c607-5df7-43d2-8365-cbbb6f2b93e1"
      },
      "source": [
        "#### Creating a custom Classifier"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "21dc0d0a-1d15-4f9e-b3ed-f83df1f51ce0",
      "metadata": {
        "id": "21dc0d0a-1d15-4f9e-b3ed-f83df1f51ce0"
      },
      "source": [
        "One thing to consider in this problem, is that the number of ALl NBA selections is fixed each year, a total of 15 players are always selected. But our current model does not limit the number of positive classified players. So, we will have to wrap our model in a custom classifier that only returns as positive the 15 players with the higher probabilities. \n",
        "Fortunately, our chosen model provides access to the underlying probabilities."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 149,
      "id": "fddf0631-d676-485d-9748-a688e854c9f2",
      "metadata": {
        "id": "fddf0631-d676-485d-9748-a688e854c9f2"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "F1:  0.769\n",
            "Dataframe of probabilities to be chosen for ALL NBA\n",
            "\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\utils\\validation.py:1858: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['str', 'tuple']. An error will be raised in 1.2.\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\utils\\validation.py:1858: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['str', 'tuple']. An error will be raised in 1.2.\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names\n",
            "  warnings.warn(\n"
          ]
        },
        {
          "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>player_id</th>\n",
              "      <th>player</th>\n",
              "      <th>PROBABILITY_OF_BEING_SELECTED</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>34</th>\n",
              "      <td>5</td>\n",
              "      <td>Alex Abrines</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>49</th>\n",
              "      <td>8</td>\n",
              "      <td>Quincy Acy</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>70</th>\n",
              "      <td>10</td>\n",
              "      <td>Jaylen Adams</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>71</th>\n",
              "      <td>13</td>\n",
              "      <td>Steven Adams</td>\n",
              "      <td>0.052</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>79</th>\n",
              "      <td>15</td>\n",
              "      <td>Bam Adebayo</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15319</th>\n",
              "      <td>2806</td>\n",
              "      <td>Trae Young</td>\n",
              "      <td>0.381</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15333</th>\n",
              "      <td>2808</td>\n",
              "      <td>Cody Zeller</td>\n",
              "      <td>0.001</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15334</th>\n",
              "      <td>2810</td>\n",
              "      <td>Tyler Zeller</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15350</th>\n",
              "      <td>2818</td>\n",
              "      <td>Ante Zizic</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15354</th>\n",
              "      <td>2820</td>\n",
              "      <td>Ivica Zubac</td>\n",
              "      <td>0.000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>530 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "       player_id        player  PROBABILITY_OF_BEING_SELECTED\n",
              "34             5  Alex Abrines                          0.000\n",
              "49             8    Quincy Acy                          0.000\n",
              "70            10  Jaylen Adams                          0.000\n",
              "71            13  Steven Adams                          0.052\n",
              "79            15   Bam Adebayo                          0.000\n",
              "...          ...           ...                            ...\n",
              "15319       2806    Trae Young                          0.381\n",
              "15333       2808   Cody Zeller                          0.001\n",
              "15334       2810  Tyler Zeller                          0.000\n",
              "15350       2818    Ante Zizic                          0.000\n",
              "15354       2820   Ivica Zubac                          0.000\n",
              "\n",
              "[530 rows x 3 columns]"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Make predictions for our dataset\n",
        "selection = best_lr.predict(all_nba_2018_df) \n",
        "\n",
        "# create the y_true (ground of truth)\n",
        "y_true = all_nba_2018_selections['all_nba'] \n",
        "\n",
        "# Let's see how well my predicton works\n",
        "print('F1: ', round(f1_score(y_true , selection),3)) \n",
        "\n",
        "# calculate the probability of being selected \n",
        "array_probs = best_lr.predict_proba(X=all_nba_2018_df)\n",
        "\n",
        "nba_full = all_nba_2018_selections.copy()\n",
        "\n",
        "print('Dataframe of probabilities to be chosen for ALL NBA')\n",
        "\n",
        "# with the calculated probability, picked the one who refers to the posibility of being selected\n",
        "nba_full['PROBABILITY_OF_BEING_SELECTED'] = list(array_probs[:, 1])\n",
        "\n",
        "# make a list with every position oredered and add it to the df\n",
        "nba_full['PREDICTIONS'] = list(selection) \n",
        "nba_full['PROBABILITY_OF_BEING_SELECTED'] = round(nba_full['PROBABILITY_OF_BEING_SELECTED'],3) \n",
        "nba_full.sort_values(by=['PROBABILITY_OF_BEING_SELECTED'], ascending=False) \n",
        "print(\"\")\n",
        "display(nba_full.loc[:,['player_id', 'player','PROBABILITY_OF_BEING_SELECTED']])\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "c1e5edbf-faee-4773-a114-eed5eab89242",
      "metadata": {
        "id": "c1e5edbf-faee-4773-a114-eed5eab89242"
      },
      "source": [
        "Sort the dataset by probabilities and show the 15 players with higher probability to have been selected"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 150,
      "id": "a99e9d7f",
      "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>player_id</th>\n",
              "      <th>player</th>\n",
              "      <th>all_nba</th>\n",
              "      <th>PROBABILITY_OF_BEING_SELECTED</th>\n",
              "      <th>PREDICTIONS</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>453</th>\n",
              "      <td>83</td>\n",
              "      <td>Giannis Antetokounmpo</td>\n",
              "      <td>1</td>\n",
              "      <td>0.999</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5604</th>\n",
              "      <td>1024</td>\n",
              "      <td>James Harden</td>\n",
              "      <td>1</td>\n",
              "      <td>0.995</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3401</th>\n",
              "      <td>587</td>\n",
              "      <td>Anthony Davis</td>\n",
              "      <td>0</td>\n",
              "      <td>0.994</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6857</th>\n",
              "      <td>1240</td>\n",
              "      <td>LeBron James</td>\n",
              "      <td>1</td>\n",
              "      <td>0.991</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4145</th>\n",
              "      <td>740</td>\n",
              "      <td>Joel Embiid</td>\n",
              "      <td>1</td>\n",
              "      <td>0.991</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3908</th>\n",
              "      <td>695</td>\n",
              "      <td>Kevin Durant</td>\n",
              "      <td>1</td>\n",
              "      <td>0.983</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14027</th>\n",
              "      <td>2572</td>\n",
              "      <td>Nikola Vucevic</td>\n",
              "      <td>0</td>\n",
              "      <td>0.972</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8206</th>\n",
              "      <td>1511</td>\n",
              "      <td>Damian Lillard</td>\n",
              "      <td>1</td>\n",
              "      <td>0.941</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14479</th>\n",
              "      <td>2648</td>\n",
              "      <td>Russell Westbrook</td>\n",
              "      <td>1</td>\n",
              "      <td>0.929</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13673</th>\n",
              "      <td>2501</td>\n",
              "      <td>Karl-Anthony Towns</td>\n",
              "      <td>0</td>\n",
              "      <td>0.920</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3168</th>\n",
              "      <td>570</td>\n",
              "      <td>Stephen Curry</td>\n",
              "      <td>1</td>\n",
              "      <td>0.901</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12473</th>\n",
              "      <td>2267</td>\n",
              "      <td>Ben Simmons</td>\n",
              "      <td>0</td>\n",
              "      <td>0.877</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6634</th>\n",
              "      <td>1204</td>\n",
              "      <td>Kyrie Irving</td>\n",
              "      <td>1</td>\n",
              "      <td>0.865</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8103</th>\n",
              "      <td>1492</td>\n",
              "      <td>Kawhi Leonard</td>\n",
              "      <td>1</td>\n",
              "      <td>0.856</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4895</th>\n",
              "      <td>884</td>\n",
              "      <td>Paul George</td>\n",
              "      <td>1</td>\n",
              "      <td>0.846</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       player_id                 player  all_nba  \\\n",
              "453           83  Giannis Antetokounmpo        1   \n",
              "5604        1024           James Harden        1   \n",
              "3401         587          Anthony Davis        0   \n",
              "6857        1240           LeBron James        1   \n",
              "4145         740            Joel Embiid        1   \n",
              "3908         695           Kevin Durant        1   \n",
              "14027       2572         Nikola Vucevic        0   \n",
              "8206        1511         Damian Lillard        1   \n",
              "14479       2648      Russell Westbrook        1   \n",
              "13673       2501     Karl-Anthony Towns        0   \n",
              "3168         570          Stephen Curry        1   \n",
              "12473       2267            Ben Simmons        0   \n",
              "6634        1204           Kyrie Irving        1   \n",
              "8103        1492          Kawhi Leonard        1   \n",
              "4895         884            Paul George        1   \n",
              "\n",
              "       PROBABILITY_OF_BEING_SELECTED  PREDICTIONS  \n",
              "453                            0.999            1  \n",
              "5604                           0.995            1  \n",
              "3401                           0.994            1  \n",
              "6857                           0.991            1  \n",
              "4145                           0.991            1  \n",
              "3908                           0.983            1  \n",
              "14027                          0.972            1  \n",
              "8206                           0.941            1  \n",
              "14479                          0.929            1  \n",
              "13673                          0.920            1  \n",
              "3168                           0.901            1  \n",
              "12473                          0.877            1  \n",
              "6634                           0.865            1  \n",
              "8103                           0.856            1  \n",
              "4895                           0.846            1  "
            ]
          },
          "execution_count": 150,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "nba_full.sort_values(by=['PROBABILITY_OF_BEING_SELECTED'], ascending=False)[0:15]"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "1c27683b-0256-40c5-929c-d5ed6cdf6d1f",
      "metadata": {
        "id": "1c27683b-0256-40c5-929c-d5ed6cdf6d1f"
      },
      "source": [
        "- Players actually selected from the all_nba_2018_selections dataset,\n",
        "- Players actually selected missing from OUR predictions+\n",
        "- Players in our predictions not selected "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 151,
      "id": "fdbcdcac-43bd-4936-8312-456b3b0bd842",
      "metadata": {
        "id": "fdbcdcac-43bd-4936-8312-456b3b0bd842"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Players actually selected from the all_nba_2018_selections dataset: \n"
          ]
        },
        {
          "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>player_id</th>\n",
              "      <th>player</th>\n",
              "      <th>all_nba</th>\n",
              "      <th>PROBABILITY_OF_BEING_SELECTED</th>\n",
              "      <th>PREDICTIONS</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>453</th>\n",
              "      <td>83</td>\n",
              "      <td>Giannis Antetokounmpo</td>\n",
              "      <td>1</td>\n",
              "      <td>0.999</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3168</th>\n",
              "      <td>570</td>\n",
              "      <td>Stephen Curry</td>\n",
              "      <td>1</td>\n",
              "      <td>0.901</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3908</th>\n",
              "      <td>695</td>\n",
              "      <td>Kevin Durant</td>\n",
              "      <td>1</td>\n",
              "      <td>0.983</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4145</th>\n",
              "      <td>740</td>\n",
              "      <td>Joel Embiid</td>\n",
              "      <td>1</td>\n",
              "      <td>0.991</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4895</th>\n",
              "      <td>884</td>\n",
              "      <td>Paul George</td>\n",
              "      <td>1</td>\n",
              "      <td>0.846</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5016</th>\n",
              "      <td>909</td>\n",
              "      <td>Rudy Gobert</td>\n",
              "      <td>1</td>\n",
              "      <td>0.822</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5399</th>\n",
              "      <td>979</td>\n",
              "      <td>Blake Griffin</td>\n",
              "      <td>1</td>\n",
              "      <td>0.665</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5604</th>\n",
              "      <td>1024</td>\n",
              "      <td>James Harden</td>\n",
              "      <td>1</td>\n",
              "      <td>0.995</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6634</th>\n",
              "      <td>1204</td>\n",
              "      <td>Kyrie Irving</td>\n",
              "      <td>1</td>\n",
              "      <td>0.865</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6857</th>\n",
              "      <td>1240</td>\n",
              "      <td>LeBron James</td>\n",
              "      <td>1</td>\n",
              "      <td>0.991</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7225</th>\n",
              "      <td>1315</td>\n",
              "      <td>Nikola Jokic</td>\n",
              "      <td>1</td>\n",
              "      <td>0.840</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8103</th>\n",
              "      <td>1492</td>\n",
              "      <td>Kawhi Leonard</td>\n",
              "      <td>1</td>\n",
              "      <td>0.856</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8206</th>\n",
              "      <td>1511</td>\n",
              "      <td>Damian Lillard</td>\n",
              "      <td>1</td>\n",
              "      <td>0.941</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14132</th>\n",
              "      <td>2585</td>\n",
              "      <td>Kemba Walker</td>\n",
              "      <td>1</td>\n",
              "      <td>0.788</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14479</th>\n",
              "      <td>2648</td>\n",
              "      <td>Russell Westbrook</td>\n",
              "      <td>1</td>\n",
              "      <td>0.929</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       player_id                 player  all_nba  \\\n",
              "453           83  Giannis Antetokounmpo        1   \n",
              "3168         570          Stephen Curry        1   \n",
              "3908         695           Kevin Durant        1   \n",
              "4145         740            Joel Embiid        1   \n",
              "4895         884            Paul George        1   \n",
              "5016         909            Rudy Gobert        1   \n",
              "5399         979          Blake Griffin        1   \n",
              "5604        1024           James Harden        1   \n",
              "6634        1204           Kyrie Irving        1   \n",
              "6857        1240           LeBron James        1   \n",
              "7225        1315           Nikola Jokic        1   \n",
              "8103        1492          Kawhi Leonard        1   \n",
              "8206        1511         Damian Lillard        1   \n",
              "14132       2585           Kemba Walker        1   \n",
              "14479       2648      Russell Westbrook        1   \n",
              "\n",
              "       PROBABILITY_OF_BEING_SELECTED  PREDICTIONS  \n",
              "453                            0.999            1  \n",
              "3168                           0.901            1  \n",
              "3908                           0.983            1  \n",
              "4145                           0.991            1  \n",
              "4895                           0.846            1  \n",
              "5016                           0.822            1  \n",
              "5399                           0.665            1  \n",
              "5604                           0.995            1  \n",
              "6634                           0.865            1  \n",
              "6857                           0.991            1  \n",
              "7225                           0.840            1  \n",
              "8103                           0.856            1  \n",
              "8206                           0.941            1  \n",
              "14132                          0.788            1  \n",
              "14479                          0.929            1  "
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Players actually selected missing from OUR predictions: \n"
          ]
        },
        {
          "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>player_id</th>\n",
              "      <th>player</th>\n",
              "      <th>all_nba</th>\n",
              "      <th>PROBABILITY_OF_BEING_SELECTED</th>\n",
              "      <th>PREDICTIONS</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "Empty DataFrame\n",
              "Columns: [player_id, player, all_nba, PROBABILITY_OF_BEING_SELECTED, PREDICTIONS]\n",
              "Index: []"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Players in our predictions not selected: \n"
          ]
        },
        {
          "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>player_id</th>\n",
              "      <th>player</th>\n",
              "      <th>all_nba</th>\n",
              "      <th>PROBABILITY_OF_BEING_SELECTED</th>\n",
              "      <th>PREDICTIONS</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>168</th>\n",
              "      <td>30</td>\n",
              "      <td>LaMarcus Aldridge</td>\n",
              "      <td>0</td>\n",
              "      <td>0.676</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1048</th>\n",
              "      <td>179</td>\n",
              "      <td>Bradley Beal</td>\n",
              "      <td>0</td>\n",
              "      <td>0.583</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3401</th>\n",
              "      <td>587</td>\n",
              "      <td>Anthony Davis</td>\n",
              "      <td>0</td>\n",
              "      <td>0.994</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3538</th>\n",
              "      <td>633</td>\n",
              "      <td>DeMar DeRozan</td>\n",
              "      <td>0</td>\n",
              "      <td>0.540</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3794</th>\n",
              "      <td>682</td>\n",
              "      <td>Andre Drummond</td>\n",
              "      <td>0</td>\n",
              "      <td>0.723</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6218</th>\n",
              "      <td>1131</td>\n",
              "      <td>Jrue Holiday</td>\n",
              "      <td>0</td>\n",
              "      <td>0.615</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12473</th>\n",
              "      <td>2267</td>\n",
              "      <td>Ben Simmons</td>\n",
              "      <td>0</td>\n",
              "      <td>0.877</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13673</th>\n",
              "      <td>2501</td>\n",
              "      <td>Karl-Anthony Towns</td>\n",
              "      <td>0</td>\n",
              "      <td>0.920</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14027</th>\n",
              "      <td>2572</td>\n",
              "      <td>Nikola Vucevic</td>\n",
              "      <td>0</td>\n",
              "      <td>0.972</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       player_id              player  all_nba  PROBABILITY_OF_BEING_SELECTED  \\\n",
              "168           30   LaMarcus Aldridge        0                          0.676   \n",
              "1048         179        Bradley Beal        0                          0.583   \n",
              "3401         587       Anthony Davis        0                          0.994   \n",
              "3538         633       DeMar DeRozan        0                          0.540   \n",
              "3794         682      Andre Drummond        0                          0.723   \n",
              "6218        1131        Jrue Holiday        0                          0.615   \n",
              "12473       2267         Ben Simmons        0                          0.877   \n",
              "13673       2501  Karl-Anthony Towns        0                          0.920   \n",
              "14027       2572      Nikola Vucevic        0                          0.972   \n",
              "\n",
              "       PREDICTIONS  \n",
              "168              1  \n",
              "1048             1  \n",
              "3401             1  \n",
              "3538             1  \n",
              "3794             1  \n",
              "6218             1  \n",
              "12473            1  \n",
              "13673            1  \n",
              "14027            1  "
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "print(\"\")\n",
        "print('Players actually selected from the all_nba_2018_selections dataset: ')\n",
        "display(nba_full[nba_full['all_nba'] == 1])\n",
        "print(\"\")\n",
        "print('Players actually selected missing from OUR predictions: ')\n",
        "display(nba_full[(nba_full['all_nba'] == 1) & (nba_full['PREDICTIONS'] == 0)])\n",
        "print(\"\")\n",
        "print('Players in our predictions not selected: ' )\n",
        "display(nba_full[(nba_full['all_nba'] == 0) & (nba_full['PREDICTIONS'] == 1)])\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "66b994d0-d37a-43c2-a38f-ea5e86afa512",
      "metadata": {
        "id": "66b994d0-d37a-43c2-a38f-ea5e86afa512"
      },
      "source": [
        "Class named AllNbaSingleSeasonClassifier that takes a scikit learn model in the constructor, then implements a predict method that returns predictions with only the 15 most probable selections."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 152,
      "id": "5fc85317-ffc5-45e9-8740-3703b5d1aa86",
      "metadata": {
        "id": "5fc85317-ffc5-45e9-8740-3703b5d1aa86"
      },
      "outputs": [],
      "source": [
        "class AllNbaSingleSeasonClassifier():\n",
        "  def __init__(self, model):\n",
        "    \"\"\"\n",
        "    Class constructor\n",
        "    \n",
        "    Parameters\n",
        "    ----------\n",
        "    model: BaseEstimator\n",
        "      a Scikit learn estimator\n",
        "    \"\"\"\n",
        "    self.model = model\n",
        "\n",
        "  def predict(self,X: pd.DataFrame) -> np.ndarray:\n",
        "    \"\"\"\n",
        "    Predicts all nba selections. This classifier will limit the number\n",
        "    of positive instances to 15\n",
        "    \n",
        "    Parameters\n",
        "    ----------\n",
        "    X: pd.DataFrame\n",
        "      a dataframe with the players data\n",
        "        \n",
        "    Returns\n",
        "    -------\n",
        "    np.ndarray\n",
        "      # An array of the same length as # y, with 15 true results based on the players with most probabilities of\n",
        "      beign selected\n",
        "    \"\"\"\n",
        "    # Predictions of my model\n",
        "    selections = self.model.predict(X)\n",
        "    probabilities = self.model.predict_proba(X)\n",
        "\n",
        "    # Round, sort and pick the first 15 better probabilities to be selected\n",
        "    probabilities = np.round(probabilities,3)\n",
        "    probabilities = probabilities[:,1]\n",
        "\n",
        "    # Copy df then convert probs & selections from arrays to columns\n",
        "    nba_best_15 = X.copy()\n",
        "    nba_best_15['PROBABILITY_OF_BEING_SELECTED'] = probabilities\n",
        "    nba_best_15['PREDICTIONS'] = selections\n",
        "\n",
        "    # sort values & put 0 to the best 1 - p probabilities of not being selected\n",
        "    nba_best_15 = nba_best_15.sort_values(by='PROBABILITY_OF_BEING_SELECTED',ascending=False)\n",
        "    nba_best_15.loc[nba_best_15.index[15:], 'PREDICTIONS'] = 0\n",
        "\n",
        "    # sort index\n",
        "    nba_best_15.sort_index(axis=0, inplace=True)\n",
        "\n",
        "    # create sorted array\n",
        "    nba_best_15_final = nba_best_15.loc[:, 'PREDICTIONS']\n",
        "\n",
        "    return nba_best_15_final\n",
        "  "
      ]
    },
    {
      "cell_type": "markdown",
      "id": "0d83848f-8132-4205-bc18-88b8a5f07555",
      "metadata": {
        "id": "0d83848f-8132-4205-bc18-88b8a5f07555"
      },
      "source": [
        "Let's use the new classifier to make predictions and evaluate the result with precision, recall and f1"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 153,
      "id": "ac440daa-8a05-4a90-8729-a48af57092bb",
      "metadata": {
        "id": "ac440daa-8a05-4a90-8729-a48af57092bb"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "\n",
            "Precision: 0.733\n",
            "\n",
            "Recall: 0.733\n",
            "\n",
            "F1: 0.733\n",
            "\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\utils\\validation.py:1858: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['str', 'tuple']. An error will be raised in 1.2.\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\utils\\validation.py:1858: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['str', 'tuple']. An error will be raised in 1.2.\n",
            "  warnings.warn(\n",
            "c:\\Users\\Fede\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\sklearn\\base.py:450: UserWarning: X does not have valid feature names, but LogisticRegression was fitted with feature names\n",
            "  warnings.warn(\n"
          ]
        }
      ],
      "source": [
        "# Create the classifier\n",
        "nba_classifier = AllNbaSingleSeasonClassifier(best_lr)\n",
        "y_pred = nba_classifier.predict(all_nba_2018_df)\n",
        "\n",
        "\n",
        "# #: Precision\n",
        "print(\"\")\n",
        "precision_my_classifier = precision_score(all_nba_2018_selections.all_nba,y_pred)\n",
        "print(f'Precision: {round(precision_my_classifier,3)}')\n",
        "print(\"\")\n",
        "# #: Recall -> The ability of the classifier to find all the POSITIVE samples. In other words, how well detects the True Positives.\n",
        "\n",
        "recall_my_classifier = recall_score(all_nba_2018_selections.all_nba,y_pred)\n",
        "print(f'Recall: {round(recall_my_classifier,3)}')\n",
        "print(\"\")\n",
        "# #: F1\n",
        "\n",
        "f1_my_classifier = f1_score(all_nba_2018_selections.all_nba,y_pred)\n",
        "print(f'F1: {round(f1_my_classifier,3)}')\n",
        "print(\"\")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "1dd382f9-9640-47ce-a077-41c8baeb2d9f",
      "metadata": {
        "id": "1dd382f9-9640-47ce-a077-41c8baeb2d9f",
        "tags": []
      },
      "source": [
        "#### Analyzing our results"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4bbefca3-c863-4cf6-a3a3-5168ef05e43f",
      "metadata": {
        "id": "4bbefca3-c863-4cf6-a3a3-5168ef05e43f"
      },
      "source": [
        "- The main problem it's the imbalaced dataset, the oversampled negative class skewed our model. Barely the 3% of our samples are from the selected ones(positive class). Our model is being biased by the negative class."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e81ac1ea-75ab-4fee-8148-20a4ea663791",
      "metadata": {
        "id": "e81ac1ea-75ab-4fee-8148-20a4ea663791"
      },
      "source": [
        "What can we do to improve our model?"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "b8f31b42-26bd-4ce4-a42d-1ceb75ba90b6",
      "metadata": {
        "id": "b8f31b42-26bd-4ce4-a42d-1ceb75ba90b6"
      },
      "source": [
        "1. Resampling: undersampling or oversampling and re-training our model\n",
        "2. Implement SMOTE to treat imbalanced datasets"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "name": "AnyoneAI - Project 2.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3.10.1 64-bit",
      "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": "3.10.7"
    },
    "vscode": {
      "interpreter": {
        "hash": "bf3fddaf5f16d859b6e8d0eeaf4cfc855ee191cae7b483f6b3cdefa4bf9da992"
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}