lithobot_train.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "lithobot_train.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd \n",
        "import numpy as np \n",
        "import os \n",
        "import pickle\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn import model_selection\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "from sklearn.datasets import make_classification\n",
        "\n",
        "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay\n",
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.metrics import classification_report\n",
        "from sklearn import metrics\n",
        "\n",
        "from sklearn.metrics import ConfusionMatrixDisplay, confusion_matrix\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "psCewTx-G-m7"
      },
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def dataloader():\n",
        "  import os \n",
        "  os.mkdir('Data')\n",
        "\n",
        "  import requests\n",
        "\n",
        "  url = 'https://github.com/bolgebrygg/Force-2020-Machine-Learning-competition/raw/master/lithology_competition/data/train.zip'\n",
        "  r = requests.get(url, allow_redirects=True)\n",
        "  print('Writing')\n",
        "  open('train.zip', 'wb').write(r.content)\n",
        "\n",
        "  print('Downloaded Zip file from XEEK server')\n",
        "\n",
        "  import zipfile\n",
        "  with zipfile.ZipFile('train.zip', 'r') as zip_ref:\n",
        "      zip_ref.extractall('Data')\n",
        "  print('Extracted in folder named Data')\n",
        "\n",
        "  import zipfile\n",
        "  with zipfile.ZipFile('train.zip', 'r') as zip_ref:\n",
        "      zip_ref.extractall('Data')\n",
        "  print('Extracted in folder named Data')\n",
        "\n",
        "  import pandas as pd \n",
        "\n",
        "  raw_data = pd.read_csv(\"Data/train.csv\", sep = ';')\n",
        "\n",
        "  lithology_keys = {30000: 'Sandstone',\n",
        "                 65030: 'Sandstone/Shale',\n",
        "                 65000: 'Shale',\n",
        "                 80000: 'Marl',\n",
        "                 74000: 'Dolomite',\n",
        "                 70000: 'Limestone',\n",
        "                 70032: 'Chalk',\n",
        "                 88000: 'Halite',\n",
        "                 86000: 'Anhydrite',\n",
        "                 99000: 'Tuff',\n",
        "                 90000: 'Coal',\n",
        "                 93000: 'Basement'}\n",
        "\n",
        "\n",
        "  raw_data = raw_data.replace({\"FORCE_2020_LITHOFACIES_LITHOLOGY\": lithology_keys})\n",
        "\n",
        "  raw_data.rename(columns={'FORCE_2020_LITHOFACIES_LITHOLOGY':'Lithofacies'}, inplace=True)\n",
        "\n",
        "  raw_data.rename(columns={'FORCE_2020_LITHOFACIES_CONFIDENCE':'Confidence'}, inplace=True)\n",
        "\n",
        "  return raw_data  "
      ],
      "metadata": {
        "id": "qJAAk8bT4tzv"
      },
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "_eFZEDmOT9D7"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "raw_data = dataloader()"
      ],
      "metadata": {
        "id": "vi7rsAA37ngc"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def thresh_u_l(raw_data):\n",
        "  # import numpy as np  \n",
        "    \n",
        "  raw_data['CALI'] = np.array(np.where((raw_data['CALI'] <= 0) | (raw_data['CALI'] >=30), np.nan, raw_data['CALI']))\n",
        "  raw_data['RSHA'] = np.array(np.where((raw_data['RSHA'] <= 0) | (raw_data['RSHA'] >=100), np.nan, raw_data['RSHA']))\n",
        "  raw_data['RMED'] = np.array(np.where((raw_data['RMED'] <= 0) | (raw_data['RMED'] >=100), np.nan, raw_data['RMED']))\n",
        "  raw_data['RDEP'] =np.array(np.where((raw_data['RDEP'] <= 0) | (raw_data['RDEP'] >= 100 ), np.nan, raw_data['RDEP']  ))\n",
        "  raw_data['RHOB'] = np.array(np.where((raw_data['RHOB'] <= 1) | (raw_data['RHOB'] >= 3), np.nan, raw_data['RHOB'])  )\n",
        "  raw_data['GR'] = np.array(np.where((raw_data['GR'] <= 0) | (raw_data['GR'] >=200), np.nan, raw_data['GR']))\n",
        "  raw_data['SGR'] = np.array(np.where((raw_data['SGR'] <= 0) | (raw_data['SGR'] >=200), np.nan, raw_data['SGR']))\n",
        "  raw_data['NPHI'] = np.array(np.where((raw_data['NPHI'] <= 0) | (raw_data['NPHI'] >= 0.6), np.nan, raw_data['NPHI'])  )\n",
        "  raw_data['PEF'] =np.array(np.where((raw_data['PEF'] <=0 ) | (raw_data['PEF'] >=10 ), np.nan, raw_data['PEF']  ))\n",
        "  raw_data['DTC'] =np.array(np.where((raw_data['DTC'] <= 0) | (raw_data['DTC'] >=300 ), np.nan, raw_data['DTC']  ))\n",
        "  raw_data['SP'] =np.array(np.where((raw_data['SP'] <= -200) | (raw_data['SP'] >= 200), np.nan, raw_data['SP']  ))\n",
        "  raw_data['BS'] =np.array(np.where((raw_data['BS'] <= 0) | (raw_data['BS'] >=30 ), np.nan, raw_data['BS']  ))\n",
        "  raw_data['ROP'] =np.array(np.where((raw_data['ROP'] <= 0) | (raw_data['ROP'] >=60 ), np.nan, raw_data['ROP']  ))\n",
        "  raw_data['DTS'] =np.array(np.where((raw_data['DTS'] <= 0) | (raw_data['DTS'] >=300 ), np.nan, raw_data['DTS']  ))\n",
        "  raw_data['DCAL'] = np.array(np.where((raw_data['DCAL'] <= 0) | (raw_data['DCAL'] >=2), np.nan, raw_data['DCAL']))\n",
        "  raw_data['DRHO'] =np.array(np.where((raw_data['DRHO'] <= -0.1) | (raw_data['DRHO'] >=0.1 ), np.nan, raw_data['DRHO']  ))\n",
        "  raw_data['MUDWEIGHT'] =np.array(np.where((raw_data['MUDWEIGHT'] <= 0) | (raw_data['MUDWEIGHT'] >=1.5 ), np.nan, raw_data['MUDWEIGHT']  ))\n",
        "  raw_data['RMIC'] =np.array(np.where((raw_data['RMIC'] <= 0) | (raw_data['RMIC'] >=30 ), np.nan, raw_data['RMIC']  ))\n",
        "  raw_data['ROPA'] =np.array(np.where((raw_data['ROPA'] <= 0) | (raw_data['ROPA'] >=60 ), np.nan, raw_data['ROPA']  ))\n",
        "  raw_data['RXO'] =np.array(np.where((raw_data['RXO'] <= 0) | (raw_data['RXO'] >=20 ), np.nan, raw_data['RXO']  ))\n",
        "\n",
        "  return raw_data\n",
        "\n",
        "def convert_nan(raw_data):\n",
        "  # import pandas as pd \n",
        "  def replace_nan(x):\n",
        "    if pd.isnull(x):\n",
        "      return 9999\n",
        "    else:\n",
        "      return x\n",
        "\n",
        "  df = pd.DataFrame()\n",
        "  for header in ['CALI', 'RSHA', 'RMED', 'RDEP', 'RHOB', 'GR', 'SGR', 'NPHI', 'PEF',        'DTC', 'SP', 'BS', 'ROP', 'DTS', 'DCAL', 'DRHO', 'MUDWEIGHT', 'RMIC',        'ROPA', 'RXO' , 'Lithofacies']: \n",
        "    df[header] = raw_data[header].apply(replace_nan)\n",
        "\n",
        "  return df\n",
        "\n",
        "def preprocess(raw_data):\n",
        "  \n",
        "  raw_data = thresh_u_l(raw_data)\n",
        "  df = convert_nan(raw_data)\n",
        "\n",
        "  return df\n"
      ],
      "metadata": {
        "id": "duBVE1zF_TSC"
      },
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df = preprocess(raw_data)"
      ],
      "metadata": {
        "id": "wdbT1FAbGHT6"
      },
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df.columns"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0PgCH3-hHQLk",
        "outputId": "4cd12620-15b6-43ed-a0f2-cbe7518f6d9e"
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['CALI', 'RSHA', 'RMED', 'RDEP', 'RHOB', 'GR', 'SGR', 'NPHI', 'PEF',\n",
              "       'DTC', 'SP', 'BS', 'ROP', 'DTS', 'DCAL', 'DRHO', 'MUDWEIGHT', 'RMIC',\n",
              "       'ROPA', 'RXO', 'Lithofacies'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {},
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def train(df,header,model_name):\n",
        "    a = 0 \n",
        "    b = 100\n",
        "    test_sz = 9.99\n",
        "\n",
        "    X_train, X_test, Y_train, Y_test = train_test_split(df[header], df['Lithofacies'], random_state = int(b),\n",
        "                                                     test_size= (test_sz/10), stratify = df['Lithofacies'])\n",
        "    clf = RandomForestClassifier(n_estimators = 200, max_depth=70,  min_samples_split=2, min_samples_leaf=1,max_features=10, random_state=0)\n",
        "    clf.fit(X_train, Y_train)\n",
        "    ## Saving model\n",
        "    if not os.path.exists('Models'):\n",
        "        os.mkdir('Models')\n",
        "    outfile = open('Models/'+model_name,'ab')\n",
        "    pickle.dump(clf,outfile)\n",
        "    outfile.close()\n",
        "\n",
        "    predicted_output = clf.predict(X_train)\n",
        "    true_lithofacies = Y_train \n",
        "\n",
        "    cm = confusion_matrix(true_lithofacies, predicted_output)\n",
        "    cmp = ConfusionMatrixDisplay(cm) \n",
        "    fig, ax = plt.subplots(figsize=(10,10))\n",
        "    cmp.plot(ax=ax)\n",
        "    plt.show()\n",
        "\n",
        "    print(\"\\nTraining Accuracy:\", accuracy_score(true_lithofacies, predicted_output))\n",
        "    print(\"\\n -------------Training Classification Report-------------\\n\")\n",
        "    print(classification_report(true_lithofacies, predicted_output))\n",
        "    \n",
        "    # predicted_output = clf.predict(X_test)\n",
        "    # true_lithofacies = Y_test \n",
        "\n",
        "    # cm = confusion_matrix(true_lithofacies, predicted_output)\n",
        "    # cmp = ConfusionMatrixDisplay(cm) # , display_labels=np.arange(25))\n",
        "    # fig, ax = plt.subplots(figsize=(10,10))\n",
        "    # cmp.plot(ax=ax)\n",
        "\n",
        "    # print(\" test Accuracy:\", accuracy_score(true_lithofacies, predicted_output))\n",
        "    # print(\"\\n ------------- Test Classification Report-------------\\n\")\n",
        "    # print(classification_report(true_lithofacies, predicted_output))\n",
        "\n",
        "\n",
        "\n",
        "def train_model1(df):\n",
        "    # print('Training started')\n",
        "    comb1 = [   'CALI', 'RHOB', 'GR', 'SGR', 'NPHI', 'PEF', 'SP', 'BS', 'DCAL', 'DRHO', 'MUDWEIGHT', 'RMIC', 'RXO' ]\n",
        "    train(df,comb1,'Model_1')"
      ],
      "metadata": {
        "id": "QTC7vU1BObew"
      },
      "execution_count": 42,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_model1(df)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 978
        },
        "id": "WkX74I9JTnha",
        "outputId": "f09e542d-d9e9-4779-b3a6-30dca9158d03"
      },
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Training Accuracy: 1.0\n",
            "\n",
            " -------------Training Classification Report-------------\n",
            "\n",
            "                 precision    recall  f1-score   support\n",
            "\n",
            "      Anhydrite       1.00      1.00      1.00         1\n",
            "          Chalk       1.00      1.00      1.00        11\n",
            "           Coal       1.00      1.00      1.00         4\n",
            "       Dolomite       1.00      1.00      1.00         2\n",
            "         Halite       1.00      1.00      1.00         8\n",
            "      Limestone       1.00      1.00      1.00        56\n",
            "           Marl       1.00      1.00      1.00        33\n",
            "      Sandstone       1.00      1.00      1.00       169\n",
            "Sandstone/Shale       1.00      1.00      1.00       150\n",
            "          Shale       1.00      1.00      1.00       721\n",
            "           Tuff       1.00      1.00      1.00        15\n",
            "\n",
            "       accuracy                           1.00      1170\n",
            "      macro avg       1.00      1.00      1.00      1170\n",
            "   weighted avg       1.00      1.00      1.00      1170\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "IwC-3XXsXzcp"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import os \n",
        "os.mkdir('Data')\n"
      ],
      "metadata": {
        "id": "Rra1IEczvnEg"
      },
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sBo9Lmy9tOBn",
        "outputId": "91fe22b7-659d-4cb1-8161-14c454ecc012"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Writing\n",
            "Downloaded\n"
          ]
        }
      ],
      "source": [
        "### Loadng data\n",
        "import requests\n",
        "\n",
        "url = 'https://github.com/bolgebrygg/Force-2020-Machine-Learning-competition/raw/master/lithology_competition/data/train.zip'\n",
        "r = requests.get(url, allow_redirects=True)\n",
        "print('Writing')\n",
        "open('train.zip', 'wb').write(r.content)\n",
        "\n",
        "print('Downloaded')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!ls"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zS_MfojsyiAF",
        "outputId": "edc33fe7-c5e4-4c9c-b825-9a88fc9875f6"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Data  sample_data  train.zip\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "VZC56kiJvctY"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import zipfile\n",
        "with zipfile.ZipFile('train.zip', 'r') as zip_ref:\n",
        "    zip_ref.extractall('Data')\n",
        "print('Extracted in folder named Data')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oAEKMHOstrct",
        "outputId": "affc13be-281a-4f8f-e865-247469b764e1"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracted\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import pandas as pd \n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "from sklearn.metrics import mean_squared_error, r2_score\n",
        "from sklearn.model_selection import train_test_split"
      ],
      "metadata": {
        "id": "rs-vLnmIty6c"
      },
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "ADM = pd.read_csv(\"Data/train.csv\", sep = ';')"
      ],
      "metadata": {
        "id": "vs8EecoQt7os"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "!ls"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "u61UVhYzy5BQ",
        "outputId": "a36e6a8a-bab8-4f37-b6b0-cc7dd0f9a9ab"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Data  sample_data  train.zip\n"
          ]
        }
      ]
    }
  ]
}