Violation_edited.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/jacobdwatters/NIOSH-Project/blob/main/Violation_edited.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "gpu_info = !nvidia-smi\n",
        "gpu_info = '\\n'.join(gpu_info)\n",
        "if gpu_info.find('failed') >= 0:\n",
        "  print('Not connected to a GPU')\n",
        "else:\n",
        "  print(gpu_info)"
      ],
      "metadata": {
        "id": "ZCCJBF_-vRvC",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4722da0e-3b5e-40c6-91ff-d8aee25805ae"
      },
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Not connected to a GPU\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from psutil import virtual_memory\n",
        "ram_gb = virtual_memory().total / 1e9\n",
        "print('Your runtime has {:.1f} gigabytes of available RAM\\n'.format(ram_gb))\n",
        "\n",
        "if ram_gb < 20:\n",
        "  print('Not using a high-RAM runtime')\n",
        "else:\n",
        "  print('You are using a high-RAM runtime!')"
      ],
      "metadata": {
        "id": "ShmmElpEtFrM",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3f777ded-e724-421a-e456-4bbd2a3316df"
      },
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Your runtime has 54.8 gigabytes of available RAM\n",
            "\n",
            "You are using a high-RAM runtime!\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 35,
      "metadata": {
        "id": "zWRkU0ZTjQ1p"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import matplotlib.pyplot as plt\n",
        "from datetime import datetime\n",
        "from sklearn.tree import plot_tree\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from sklearn.model_selection import cross_val_score\n",
        "from sklearn_pandas import DataFrameMapper\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.metrics import plot_confusion_matrix\n",
        "from sklearn.preprocessing import OneHotEncoder\n",
        "import xgboost as xgb\n",
        "from sklearn import set_config\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from joblib import dump, load\n",
        "\n",
        "pd.set_option('display.max_rows', 100)\n",
        "pd.set_option('display.max_columns', 100)\n",
        "pd.set_option('display.width', 1000)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "metadata": {
        "id": "z1k9w52YwHDH",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "991e2931-7c6b-44f6-db62-e02cdeff820b"
      },
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "id": "cjTPHF9qjbNw"
      },
      "outputs": [],
      "source": [
        "# Import Datasets and PreProcessing\n",
        "\n",
        "# violations = pd.read_csv (\"https://arlweb.msha.gov/OpenGovernmentData/DataSets/Violations.zip\", encoding='latin-1', compression='zip', sep='|')\n",
        "\n",
        "# violations['VIOLATOR_VIOLATION_CNT'] = violations['VIOLATOR_VIOLATION_CNT'].fillna('0')\n",
        "# violations['PROPOSED_PENALTY'] = violations['PROPOSED_PENALTY'].fillna('0')\n",
        "# violations['MINE_TYPE'] = violations['MINE_TYPE'].fillna('Facility')\n",
        "# violations['LIKELIHOOD'] = violations['LIKELIHOOD'].fillna('NoLikelihood')\n",
        "# violations['PROPOSED_PENALTY'] = violations['PROPOSED_PENALTY'].fillna('0')\n",
        "# violations['NEGLIGENCE'] = violations['NEGLIGENCE'].fillna('NONEGLIGENCE')\n",
        "# violations['NEGLIGENCE'] = violations['VIOLATION_ISSUE_TIME'].fillna('2319')\n",
        "# violations['COAL_METAL_IND'] = violations['COAL_METAL_IND'].fillna('M')\n",
        "# violations['NO_AFFECTED'] = violations['NO_AFFECTED'].fillna('0')\n",
        "# violations['INJ_ILLNESS'] = violations['INJ_ILLNESS'].fillna('NoLostDays')\n",
        "# violations['SIG_SUB'] = violations['SIG_SUB'].fillna('N')\n",
        "\n",
        "# violations['VIOLATION_OCCUR_DT'] = pd.to_datetime(violations['VIOLATION_OCCUR_DT'], format='%m/%d/%Y', exact=False)\n",
        "# violations.reset_index(inplace=True)\n",
        "# violations['day_occur'] = violations['VIOLATION_OCCUR_DT'].dt.day\n",
        "# violations['month_occur'] = violations['VIOLATION_OCCUR_DT'].dt.month\n",
        "# violations['year_occur'] = violations['VIOLATION_OCCUR_DT'].dt.year\n",
        "\n",
        "# violations['year_occur'] = violations['year_occur'].fillna('1999')\n",
        "\n",
        "# violations = violations._convert(numeric=True)\n",
        "\n",
        "violations = pd.read_csv('gdrive/My Drive/NIOSH Project/data/violations_processed.csv')\n",
        "\n",
        "# Selecting Parameters\n",
        "X = violations[['year_occur', \n",
        "          'VIOLATION_ISSUE_TIME',\n",
        "          'VIOLATOR_VIOLATION_CNT',\n",
        "          'MINE_TYPE', \n",
        "          'COAL_METAL_IND', \n",
        "          'NEGLIGENCE',\n",
        "          'LIKELIHOOD', \n",
        "          'NO_AFFECTED',\n",
        "          'INJ_ILLNESS',\n",
        "          'SIG_SUB']]\n",
        "\n",
        "y = violations['PROPOSED_PENALTY']\n",
        "\n",
        "# processed_data = pd.concat([X, y], axis=1)\n",
        "# path = '/content/gdrive/My Drive/NIOSH Project/data/violations_processed.csv'\n",
        "# with open(path, 'w', encoding='utf-8') as f:\n",
        "#   processed_data.to_csv(f)\n",
        "\n",
        "# del(processed_data)\n",
        "del(violations)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 38,
      "metadata": {
        "id": "bexZ16vrk3Ad",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "317ac004-b248-4209-c1f4-40ee0819aa0b"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "year_occur                0\n",
              "VIOLATION_ISSUE_TIME      1\n",
              "VIOLATOR_VIOLATION_CNT    0\n",
              "MINE_TYPE                 0\n",
              "COAL_METAL_IND            0\n",
              "NEGLIGENCE                0\n",
              "LIKELIHOOD                0\n",
              "NO_AFFECTED               0\n",
              "INJ_ILLNESS               0\n",
              "SIG_SUB                   0\n",
              "dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 38
        }
      ],
      "source": [
        "X.isnull().sum()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 39,
      "metadata": {
        "id": "KrfX_ZCkqAyW",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c4dfb603-20aa-49ac-982d-252c16229958"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "N    1998484\n",
              "Y     755523\n",
              "Name: SIG_SUB, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ],
      "source": [
        "X.SIG_SUB.value_counts(dropna=False)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 40,
      "metadata": {
        "id": "tTApUygHsEkA",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "50c6f871-6e6f-4241-ecd7-a43be6843cae"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "year_occur                float64\n",
              "VIOLATION_ISSUE_TIME      float64\n",
              "VIOLATOR_VIOLATION_CNT    float64\n",
              "MINE_TYPE                  object\n",
              "COAL_METAL_IND             object\n",
              "NEGLIGENCE                float64\n",
              "LIKELIHOOD                 object\n",
              "NO_AFFECTED               float64\n",
              "INJ_ILLNESS                object\n",
              "SIG_SUB                    object\n",
              "dtype: object"
            ]
          },
          "metadata": {},
          "execution_count": 40
        }
      ],
      "source": [
        "X.dtypes"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 41,
      "metadata": {
        "id": "UkMuX7itjpg9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0ed567b4-507b-45b7-fa78-2503c2ceba1d"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Features shape: (2754007, 21)\n",
            "No.of.unique values in y: 1957\n"
          ]
        }
      ],
      "source": [
        "# Define which columns should be encoded vs scaled\n",
        "columns_to_encode = ['MINE_TYPE','COAL_METAL_IND','LIKELIHOOD','INJ_ILLNESS','SIG_SUB']\n",
        "columns_to_scale  = ['year_occur', 'VIOLATION_ISSUE_TIME', 'NEGLIGENCE', 'VIOLATOR_VIOLATION_CNT','NO_AFFECTED']\n",
        "\n",
        "# Instantiate encoder/scaler\n",
        "scaler = StandardScaler()\n",
        "ohe = OneHotEncoder(sparse=False)\n",
        "\n",
        "# Scale and Encode Separate Columns\n",
        "scaled_columns  = scaler.fit_transform(X[columns_to_scale])\n",
        "encoded_columns = ohe.fit_transform(X[columns_to_encode])\n",
        "\n",
        "# Concatenate (Column-Bind) Processed Columns Back Together\n",
        "X_pre = np.concatenate([scaled_columns, encoded_columns], axis=1)\n",
        "np.nan_to_num(X_pre, copy=False)\n",
        "\n",
        "print(\"Features shape:\", X_pre.shape)\n",
        "\n",
        "\n",
        "# Bin the values into 5 equally sized bins.\n",
        "y_bins = pd.cut(y, 5)\n",
        "\n",
        "le = LabelEncoder()\n",
        "y_pre = le.fit_transform(y_bins)\n",
        "\n",
        "n = len(pd.unique(y))\n",
        "print(\"No.of.unique values in y:\", n)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 42,
      "metadata": {
        "id": "muy375z1jsRo"
      },
      "outputs": [],
      "source": [
        "# Splitting the dataset into the Training set and Test set\n",
        "X_train, X_test, y_train, y_test = train_test_split(X_pre, y_pre, test_size = 0.25, random_state = 0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 43,
      "metadata": {
        "id": "kWTCxDYSjutN",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1e010d3e-fcfa-4cfd-f666-19b4311c87ab"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy of logistic regression: 0.999\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['gdrive/My Drive/NIOSH Project/violation_logreg_trained.joblib']"
            ]
          },
          "metadata": {},
          "execution_count": 43
        }
      ],
      "source": [
        "# Fitting classifier to the Training set\n",
        "set_config(display='diagram')\n",
        "\n",
        "model = LogisticRegression(max_iter=300000)\n",
        "model.fit(X_train, y_train)\n",
        "\n",
        "accuracy = model.score(X_test, y_test)\n",
        "print(f\"Accuracy of logistic regression: {accuracy:.3f}\")\n",
        "\n",
        "# Save the model\n",
        "dump(model, 'gdrive/My Drive/NIOSH Project/violation_logreg_trained.joblib') "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 44,
      "metadata": {
        "id": "MJla7fnYjxd3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 656
        },
        "outputId": "43136e5c-3563-40e2-b2e3-15170480c29a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function plot_confusion_matrix is deprecated; Function `plot_confusion_matrix` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: ConfusionMatrixDisplay.from_predictions or ConfusionMatrixDisplay.from_estimator.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7f94b38807d0>"
            ]
          },
          "metadata": {},
          "execution_count": 44
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "fig, ax = plt.subplots(figsize=(10, 10))\n",
        "plot_confusion_matrix(model, X_test, y_test, ax=ax, normalize='true')\n",
        "plt.title('Normalized Confusion Matrix')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "fig, ax = plt.subplots(figsize=(10, 10))\n",
        "plot_confusion_matrix(model, X_test, y_test, ax=ax)\n",
        "plt.title('Confusion Matrix')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 652
        },
        "id": "kkHVmazrxTZ7",
        "outputId": "332607a4-b8f0-47fc-e40c-8f4680561488"
      },
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function plot_confusion_matrix is deprecated; Function `plot_confusion_matrix` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: ConfusionMatrixDisplay.from_predictions or ConfusionMatrixDisplay.from_estimator.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Confusion Matrix')"
            ]
          },
          "metadata": {},
          "execution_count": 46
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "provenance": [],
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    },
    "gpuClass": "premium"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}