Ensemble_Models.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "CA04 - Ensemble Models",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gQg5tR_pmY5W"
      },
      "source": [
        "## 1. Data Source and Contents\n",
        "\n",
        "The dataset is obtained from the Census Bureau and represents salaries of people along with seven demographic variables. The following is a description of our dataset:\n",
        "\n",
        "* **Number of target classes:** 2 ('>50K' and '<=50K') [ Labels: 1, 0 ]\n",
        "* **Number of attributes (Columns):** 7\n",
        "* **Number of instances (Rows):** 48,842"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4RAPdP8W7m0_",
        "outputId": "d073f0bb-7732-45fb-f38b-b5ae804cfee3"
      },
      "source": [
        "# importing packages\n",
        "!pip install xgboost\n",
        "\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from sklearn.tree import DecisionTreeClassifier\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.ensemble import AdaBoostClassifier\n",
        "from sklearn.ensemble import GradientBoostingClassifier\n",
        "from xgboost import XGBClassifier\n",
        "from sklearn.metrics import roc_auc_score"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: xgboost in /usr/local/lib/python3.7/dist-packages (0.90)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from xgboost) (1.19.5)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.7/dist-packages (from xgboost) (1.4.1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hZZXK4jsySJR",
        "outputId": "f2376603-d79e-406f-e4e2-31a43833f886"
      },
      "source": [
        "# mounting drive\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3oGs_owYnBuT"
      },
      "source": [
        "# reading the data\n",
        "census_data = pd.read_csv('/content/drive/MyDrive/MSBA_Colab_2020/ML_Algorithms/CA04/Data/census_data.csv')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "8jW8okLqnZb4",
        "outputId": "87c2c4ea-b065-40b5-e444-9324642f508a"
      },
      "source": [
        "# checking out the dataset\n",
        "census_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>hours_per_week_bin</th>\n",
              "      <th>occupation_bin</th>\n",
              "      <th>msr_bin</th>\n",
              "      <th>capital_gl_bin</th>\n",
              "      <th>race_sex_bin</th>\n",
              "      <th>education_num_bin</th>\n",
              "      <th>education_bin</th>\n",
              "      <th>workclass_bin</th>\n",
              "      <th>age_bin</th>\n",
              "      <th>flag</th>\n",
              "      <th>y</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>b. 31-40</td>\n",
              "      <td>b. Mid - Low</td>\n",
              "      <td>b. Mid</td>\n",
              "      <td>c. &gt; 0</td>\n",
              "      <td>c. High</td>\n",
              "      <td>c. 13</td>\n",
              "      <td>c. Bachelors</td>\n",
              "      <td>b. income</td>\n",
              "      <td>d. 36-40 &amp; 56-60</td>\n",
              "      <td>train</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>a. 0-30</td>\n",
              "      <td>e. High</td>\n",
              "      <td>c. High</td>\n",
              "      <td>a. = 0</td>\n",
              "      <td>c. High</td>\n",
              "      <td>c. 13</td>\n",
              "      <td>c. Bachelors</td>\n",
              "      <td>b. income</td>\n",
              "      <td>e. 40-55</td>\n",
              "      <td>train</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>b. 31-40</td>\n",
              "      <td>a. Low</td>\n",
              "      <td>b. Mid</td>\n",
              "      <td>a. = 0</td>\n",
              "      <td>c. High</td>\n",
              "      <td>b. 9-12</td>\n",
              "      <td>b. Mid</td>\n",
              "      <td>b. income</td>\n",
              "      <td>d. 36-40 &amp; 56-60</td>\n",
              "      <td>train</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>b. 31-40</td>\n",
              "      <td>a. Low</td>\n",
              "      <td>c. High</td>\n",
              "      <td>a. = 0</td>\n",
              "      <td>b. Mid</td>\n",
              "      <td>a. 0-8</td>\n",
              "      <td>a. Low</td>\n",
              "      <td>b. income</td>\n",
              "      <td>e. 40-55</td>\n",
              "      <td>train</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>b. 31-40</td>\n",
              "      <td>e. High</td>\n",
              "      <td>c. High</td>\n",
              "      <td>a. = 0</td>\n",
              "      <td>a. Low</td>\n",
              "      <td>c. 13</td>\n",
              "      <td>c. Bachelors</td>\n",
              "      <td>b. income</td>\n",
              "      <td>b. 26-30 &amp; 71-100</td>\n",
              "      <td>train</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "  hours_per_week_bin occupation_bin  msr_bin  ...            age_bin   flag  y\n",
              "0           b. 31-40   b. Mid - Low   b. Mid  ...   d. 36-40 & 56-60  train  0\n",
              "1            a. 0-30        e. High  c. High  ...           e. 40-55  train  0\n",
              "2           b. 31-40         a. Low   b. Mid  ...   d. 36-40 & 56-60  train  0\n",
              "3           b. 31-40         a. Low  c. High  ...           e. 40-55  train  0\n",
              "4           b. 31-40        e. High  c. High  ...  b. 26-30 & 71-100  train  0\n",
              "\n",
              "[5 rows x 11 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 177
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q3byfcu4mpXg"
      },
      "source": [
        "## 2. Finding Optimal Value of a Key Hyper-parameter\n",
        "For Ensemble Models, one of the key hyper-parameter is number of “estimators”. The following is an example of finding the optimal value of “Maximum Depth”. Review the following code snippet to understand how the optimal value of this hyper-parameter (Max Depth = 10) is found by plotting a graph."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VRoU10rPpGyS",
        "outputId": "39944d39-1a02-46f5-8728-6488a3c27bc0"
      },
      "source": [
        "# encoding categorical values\n",
        "census_data_encoded = census_data.apply(LabelEncoder().fit_transform)\n",
        "census_data_encoded.dtypes\n",
        "\n",
        "# flag: train = 1, test = 0\n",
        "\n",
        "# seperating dataset into training (67%) and testing (33%) data\n",
        "X_train = census_data_encoded[census_data_encoded.flag==1].drop(['flag','y'], axis=1)\n",
        "X_test = census_data_encoded[census_data_encoded.flag==0].drop(['flag','y'], axis=1).reset_index(drop=True)\n",
        "\n",
        "y_train = census_data_encoded[census_data_encoded.flag==1][['y']]\n",
        "y_test = census_data_encoded[census_data_encoded.flag==0][['y']].reset_index(drop=True)\n",
        "\n",
        "# verifying shapes (split ratio)\n",
        "X_train.shape\n",
        "X_test.shape\n",
        "y_train.shape\n",
        "y_test.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(16281, 1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 178
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2MV3XDHv7wbt",
        "outputId": "456a5ac9-f3b4-44d1-f0cb-6c4d59045671"
      },
      "source": [
        "# verifying that all categorical data has been transformed into numerical data\n",
        "census_data_encoded.info()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 48842 entries, 0 to 48841\n",
            "Data columns (total 11 columns):\n",
            " #   Column              Non-Null Count  Dtype\n",
            "---  ------              --------------  -----\n",
            " 0   hours_per_week_bin  48842 non-null  int64\n",
            " 1   occupation_bin      48842 non-null  int64\n",
            " 2   msr_bin             48842 non-null  int64\n",
            " 3   capital_gl_bin      48842 non-null  int64\n",
            " 4   race_sex_bin        48842 non-null  int64\n",
            " 5   education_num_bin   48842 non-null  int64\n",
            " 6   education_bin       48842 non-null  int64\n",
            " 7   workclass_bin       48842 non-null  int64\n",
            " 8   age_bin             48842 non-null  int64\n",
            " 9   flag                48842 non-null  int64\n",
            " 10  y                   48842 non-null  int64\n",
            "dtypes: int64(11)\n",
            "memory usage: 4.1 MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dmpoGMVflIX0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "outputId": "e9822e5a-a74d-44c8-f1a0-162d4b6ba68a"
      },
      "source": [
        "# calculating accuracy score for each maximum depth value to find optimal max_depth\n",
        "\n",
        "results = []\n",
        "max_depth_options = [2,4,6,8,10,12,14,16,18,20] # different options for maximum tree depths \n",
        "\n",
        "for trees in max_depth_options: # iterates through every maximum depth value\n",
        "  model = DecisionTreeClassifier(max_depth=trees, random_state=101) # creates Decision Tree model for each maximum depth value \n",
        "  model.fit(X_train, y_train) # fits model with training data defined above\n",
        "  y_pred = model.predict(X_test)  # predicting outcomes (y) using outcome features of test data\n",
        "  accuracy = accuracy_score(y_test, y_pred) # calculating accuracy score of prediction (used in palce of np.mean)\n",
        "  results.append(accuracy)  # appends accuracy scores to empty results list \n",
        "\n",
        "\n",
        "#plotting result\n",
        "\n",
        "plt.figure(figsize=(8,4))\n",
        "pd.Series(results, max_depth_options).plot(color='darkred', marker='o')\n",
        "plt.title('Decision Tree Model: Accuracy vs Optimal Maximum Depth')\n",
        "plt.xlabel('Maximum Depth')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oZStl3SI3m9D"
      },
      "source": [
        "## 3. Building a Random Forest Model\n",
        "Using Notebook, and the same data source from CA03, train a Random Forest Model. Using similar approach of Section 2 above, plot a graph of Accuracy vs. n_estimator. Use n_estimator values as [50,100,150,200,250,300,350,400,450,500]."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "0sIso9-u33wf",
        "outputId": "72408b1f-f5de-4ba7-e2fc-e94e180d210d"
      },
      "source": [
        "# calculating accuracy score for each n_estimator value to find optimal n_estimator\n",
        "\n",
        "n_estimator_options = [50,100,150,200,250,300,350,400,450,500] # n_estimator values\n",
        "results = []\n",
        "\n",
        "for trees in n_estimator_options:  # iterates through list of n_values\n",
        "  model = RandomForestClassifier(n_estimators=trees, max_features='auto', random_state=101) # applies Random Forest model to each n_estimator value. Changing shape of y to match n_samples using .ravel()\n",
        "  model.fit(X_train, y_train.values.ravel()) # fits model using train and test data defined above. \n",
        "  y_pred = model.predict(X_test)  # predicting outcomes (y) using outcome features of test data\n",
        "  accuracy = accuracy_score(y_test, y_pred) # calculating accuracy score of prediction\n",
        "  results.append(accuracy)  # appends accuracy score of each n_estimator value to results list\n",
        "\n",
        "\n",
        "# plotting result\n",
        "\n",
        "plt.figure(figsize=(8,4))\n",
        "pd.Series(results, n_estimator_options).plot(color='darkblue', marker = 'o')\n",
        "plt.title('Random Forest Model: Accuracy vs n_estimator')\n",
        "plt.xlabel('n_estimator')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7t_Q1O9oFVMz"
      },
      "source": [
        "Answer the following question:\n",
        "1. **Write your observations about the Classifier’s behavior with respect to the\n",
        "number of estimators**\n",
        "\n",
        "> The plot returns a polynomial trendline. The accuracy score of the predictions gradually decreases when the n_estimator ranges between 50 and 250 trees. At 300 trees, the accuracy increases again before decreasing at 350. After, it gradually increases again until it reaches 500 trees. Thus, the model that uses the lowest n_estimator value produces the optimal results in terms of accuracy.\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "2. **Is there an optimal value of the estimator within the given range?**\n",
        "\n",
        "> The optimal value of the estimator is 50 trees, which results in the highest accuracy score of 0.8384."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oYSTmxabIkmy"
      },
      "source": [
        "## 4. AdaBoost Model\n",
        "Using a similar process as in Section 3, build an AdaBoost Model and plot a graph of Accuracy vs n_estimator."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "Q3_qAWWlI4rG",
        "outputId": "b8950171-bf0e-4498-eecc-1a64553192f5"
      },
      "source": [
        "# calculating accuracy score for each n_estimator value to find optimal n_estimator\n",
        "\n",
        "n_estimator_options = [50,100,150,200,250,300,350,400,450,500] # n_estimator values\n",
        "results = []\n",
        "\n",
        "for trees in n_estimator_options:  # iterates through list of n_values\n",
        "  model = AdaBoostClassifier(n_estimators=trees, random_state=101) # applies AdaBoost model to each n_estimator value. Changing shape of y to match n_samples using .ravel()\n",
        "  model.fit(X_train, y_train.values.ravel()) # fits model using train and test data defined above. \n",
        "  y_pred = model.predict(X_test)  # predicting outcomes (y) using outcome features of test data\n",
        "  accuracy = accuracy_score(y_test, y_pred) # calculating accuracy score of prediction\n",
        "  results.append(accuracy)  # appends accuracy score of each n_estimator value to results list\n",
        "\n",
        "\n",
        "# plotting result\n",
        "\n",
        "plt.figure(figsize=(8,4))\n",
        "pd.Series(results, n_estimator_options).plot(color='darkgreen', marker = 'o')\n",
        "plt.title('AdaBoost Model: Accuracy vs n_estimator')\n",
        "plt.xlabel('n_estimator')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BSxC6M2gJ_QU"
      },
      "source": [
        "1. **Write your observations about the Classifier’s behavior with respect to the\n",
        "number of estimators**\n",
        "\n",
        "> The accuracy score starts at a low with 50 trees, but then peaks with 100-150 trees. It decreases again once it reaches 200, and stays constant until 500 trees. To obtain the optimal results, a relatively low n_estimator value should be used. Note that there is an improvement in the overall accuracy scores compared to the Random Forest Model.\n",
        "\n",
        "2. **Is there an optimal value of the estimator within the given range?**\n",
        "\n",
        "> The optimal value of the estimator is 100-150 trees, which results in the highest accuracy scores of 0.844915."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RVg82LESKxIq"
      },
      "source": [
        "## 4. Gradient Boost Model (classifier)\n",
        "Using a similar process as in Section 3, build an Gradient Boost Model and plot a graph of Accuracy vs n_estimator."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "NEuyDEVHLMWi",
        "outputId": "9472a8e3-7ea1-484c-d63f-bd51d74752df"
      },
      "source": [
        "# calculating accuracy score for each n_estimator value to find optimal n_estimator\n",
        "\n",
        "n_estimator_options = [50,100,150,200,250,300,350,400,450,500] # n_estimator values\n",
        "results = []\n",
        "\n",
        "for trees in n_estimator_options:  # iterates through list of n_values\n",
        "  model = GradientBoostingClassifier(n_estimators=trees, random_state=101) # applies Gradient Boost model to each n_estimator value. Changing shape of y to match n_samples using .ravel()\n",
        "  model.fit(X_train, y_train.values.ravel()) # fits model using train and test data defined above. \n",
        "  y_pred = model.predict(X_test)  # predicting outcomes (y) using outcome features of test data\n",
        "  accuracy = accuracy_score(y_test, y_pred) # calculating accuracy score of prediction\n",
        "  results.append(accuracy)  # appends accuracy score of each n_estimator value to results list\n",
        "\n",
        "\n",
        "# plotting result\n",
        "\n",
        "plt.figure(figsize=(8,4))\n",
        "pd.Series(results, n_estimator_options).plot(color='darkorange', marker = 'o')\n",
        "plt.title('Gradient Boost Model: Accuracy vs n_estimator')\n",
        "plt.xlabel('n_estimator')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wGyKF5TWL0vv"
      },
      "source": [
        "1. **Write your observations about the Classifier’s behavior with respect to the\n",
        "number of estimators**\n",
        "\n",
        "> The graph returned a polynomial trendline. With 50 trees, the accuracy score is relatively high but then decreases within the 100-200 tree range. In the range of 250-500 trees, the trendline fluctuates heaviliy but reaches peak accuracy at 350 trees. Thus, a medium-high n_estimator value should be used to obtain optimal results. Note that the accuracy score using the Gradient Boost model is higher than that using the AdaBoost or Random Forest model.\n",
        "\n",
        "2. **Is there an optimal value of the estimator within the given range?**\n",
        "\n",
        "> The optimal value of the estimator is 350 trees, which results in the highest accuracy scores of 0.8472."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QIjVgA2jNQGC"
      },
      "source": [
        "## 4. XGBoost Model\n",
        "Using a similar process as in Section 3, build an Gradient Boost Model and plot a graph of Accuracy vs n_estimator.\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "xmYZ5gV_OIEB",
        "outputId": "0e06b11e-7456-4fa9-c8ce-fc866578b64f"
      },
      "source": [
        "# calculating accuracy score for each n_estimator value to find optimal n_estimator\n",
        "\n",
        "n_estimator_options = [50,100,150,200,250,300,350,400,450,500] # n_estimator values\n",
        "results = []\n",
        "\n",
        "for trees in n_estimator_options:  # iterates through list of n_values\n",
        "  model = XGBClassifier(n_estimators=trees, random_state=101) # applies XGBoost Model to each n_estimator value. Changing shape of y to match n_samples using .ravel()\n",
        "  model.fit(X_train, y_train.values.ravel()) # fits model using train and test data defined above. \n",
        "  y_pred = model.predict(X_test)  # predicting outcomes (y) using outcome features of test data\n",
        "  accuracy = accuracy_score(y_test, y_pred) # calculating accuracy score of prediction\n",
        "  results.append(accuracy)  # appends accuracy score of each n_estimator value to results list\n",
        "\n",
        "\n",
        "# plotting result\n",
        "\n",
        "plt.figure(figsize=(8,4))\n",
        "pd.Series(results, n_estimator_options).plot(color='indigo', marker = 'o')\n",
        "plt.title('XGBoost Model: Accuracy vs n_estimator')\n",
        "plt.xlabel('n_estimator')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.grid()\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xv9nV3XROoio"
      },
      "source": [
        "1. **Write your observations about the Classifier’s behavior with respect to thenumber of estimators**\n",
        "\n",
        "> The graph returned a trendline that follows a non-linear downward trend. It returns the highest accuracy score at 100 trees. Despite there are small increases in accuracy, it gradually decreases until it reaches 500 trees. \n",
        "Thus, a relatively low n_estimator value should be used to obtain optimal results. The range of accuracy scores is comparable to those when using the Gradient Boost Model.\n",
        "\n",
        "2. **Is there an optimal value of the estimator within the given range?**\n",
        "\n",
        "> The optimal value of the estimator is 100 trees, which results in the highest accuracy scores of about 0.84715."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CGsdPcZCPjIT"
      },
      "source": [
        "## 5. Compare Performance\n",
        "Keep all common Hyper-parameters same for four models (Random Forest, AdaBoost, Gradient Boost, XGB), run them again and create a performance comparison table within your.\n",
        "Also, print the values of all COMMON Hyper-parameters below this table (from code)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EhOA8e_zT2Rt",
        "outputId": "933bc11b-a817-4bf3-fda8-dabbf15cfe04"
      },
      "source": [
        "# re-calculating accuracy score with n_estimator = 250\n",
        "n_value = 250\n",
        "accuracy_results = []\n",
        "\n",
        "# calculating values and appending results to list\n",
        "\n",
        "# random forest\n",
        "model_rf = RandomForestClassifier(n_estimators=n_value, max_features='auto', random_state=101)\n",
        "model_rf.fit(X_train, y_train.values.ravel())\n",
        "y_pred_rf = model_rf.predict(X_test)\n",
        "accuracy_results.append(accuracy_score(y_test, y_pred_rf))\n",
        "\n",
        "# ada boost\n",
        "model_ada = AdaBoostClassifier(n_estimators=n_value, random_state=101)\n",
        "model_ada.fit(X_train, y_train.values.ravel())\n",
        "y_pred_ada = model_ada.predict(X_test)\n",
        "accuracy_results.append(accuracy_score(y_test, y_pred_ada))\n",
        "\n",
        "# gradient boost\n",
        "model_gb = GradientBoostingClassifier(n_estimators=n_value, random_state=101)\n",
        "model_gb.fit(X_train, y_train.values.ravel())\n",
        "y_pred_gb = model_gb.predict(X_test)\n",
        "accuracy_results.append(accuracy_score(y_test, y_pred_gb))\n",
        "\n",
        "# XGB\n",
        "model_xgb = XGBClassifier(n_estimators=n_value, random_state=101)\n",
        "model_xgb.fit(X_train, y_train.values.ravel())\n",
        "y_pred_xgb = model_xgb.predict(X_test)\n",
        "accuracy_results.append(accuracy_score(y_test, y_pred_xgb))\n",
        "\n",
        "accuracy_results"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.8370493212947607, 0.8447884036607088, 0.847122412628217, 0.8467538848965052]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 185
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FUlgWFGJaB2C",
        "outputId": "bba40563-3e9f-47d3-dafd-4128190317ce"
      },
      "source": [
        "# calculating AUC score with n_estimator = 250\n",
        "auc_results = []\n",
        "\n",
        "# calculating values and appending results to list\n",
        "auc_results.append(roc_auc_score(y_test, y_pred_rf))\n",
        "auc_results.append(roc_auc_score(y_test, y_pred_ada))\n",
        "auc_results.append(roc_auc_score(y_test, y_pred_gb))\n",
        "auc_results.append(roc_auc_score(y_test, y_pred_xgb))\n",
        "\n",
        "auc_results"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.7450719090283514,\n",
              " 0.7460076328264228,\n",
              " 0.7598376456168019,\n",
              " 0.7597759833191879]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 186
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 145
        },
        "id": "mVFSNyBXP39U",
        "outputId": "e1e1e86c-7e2f-4a09-8c45-b45cc96025c7"
      },
      "source": [
        "results = []\n",
        "results.append([accuracy_results, auc_results])\n",
        "\n",
        "# creating empty table\n",
        "performance = pd.DataFrame(index=['Accuracy','AUC'],columns=['Random Forst','AdaBoost','Gradient Boost','XGB'])\n",
        "performance.loc['Accuracy'] = accuracy_results\n",
        "performance.loc['AUC'] = auc_results\n",
        "\n",
        "# find common parameters\n",
        "rf_hp = model_rf.get_params()\n",
        "ada_hp = model_ada.get_params()\n",
        "gb_hp = model_gb.get_params()\n",
        "xgb_hp = model_xgb.get_params()\n",
        "\n",
        "common_hp = rf_hp.keys() & ada_hp.keys() & gb_hp.keys() & xgb_hp.keys()\n",
        "\n",
        "display(performance)\n",
        "\n",
        "print('\\nThe common hyperparameters are ', common_hp)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "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>Random Forst</th>\n",
              "      <th>AdaBoost</th>\n",
              "      <th>Gradient Boost</th>\n",
              "      <th>XGB</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Accuracy</th>\n",
              "      <td>0.837049</td>\n",
              "      <td>0.844788</td>\n",
              "      <td>0.847122</td>\n",
              "      <td>0.846754</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>AUC</th>\n",
              "      <td>0.745072</td>\n",
              "      <td>0.746008</td>\n",
              "      <td>0.759838</td>\n",
              "      <td>0.759776</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         Random Forst  AdaBoost Gradient Boost       XGB\n",
              "Accuracy     0.837049  0.844788       0.847122  0.846754\n",
              "AUC          0.745072  0.746008       0.759838  0.759776"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "\n",
            "The common hyperparameters are  {'random_state', 'n_estimators'}\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}