scikit_learn_with_mlflow.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "mount_file_id": "16-7nCK4EjEeqRAvwkT1TQGqFGefMG71m",
      "authorship_tag": "ABX9TyMdfwpb1YdQRKMUMpwwsZBV",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/workdd/MLOps_with_MLFlow/blob/main/scikit_learn_with_mlflow.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# MLFlow를 활용한 MLOps 챕터4"
      ],
      "metadata": {
        "id": "eEID_4t-iIZH"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# 사이킷런을 활용한 MLFlow (p124)"
      ],
      "metadata": {
        "id": "SUOUWGItzARq"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 필수 라이브러리 설치 및 버전 확인"
      ],
      "metadata": {
        "id": "bYK_h4Hm23da"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install mlflow==1.10.0\n",
        "#scikit-learn 1.2 버전 아래에서 진행해야 원활히 진행 가능\n",
        "!pip install scikit-learn==1.1.0"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0uZDKbHx5dnM",
        "outputId": "102b4b47-3370-461b-8987-56c245cd04aa"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting mlflow==1.10.0\n",
            "  Downloading mlflow-1.10.0-py3-none-any.whl (12.4 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m12.4/12.4 MB\u001b[0m \u001b[31m92.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting alembic (from mlflow==1.10.0)\n",
            "  Downloading alembic-1.11.1-py3-none-any.whl (224 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m224.5/224.5 kB\u001b[0m \u001b[31m26.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting azure-storage-blob>=12.0 (from mlflow==1.10.0)\n",
            "  Downloading azure_storage_blob-12.16.0-py3-none-any.whl (387 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m388.0/388.0 kB\u001b[0m \u001b[31m36.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: click>=7.0 in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (8.1.3)\n",
            "Requirement already satisfied: cloudpickle in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (2.2.1)\n",
            "Collecting databricks-cli>=0.8.7 (from mlflow==1.10.0)\n",
            "  Downloading databricks-cli-0.17.7.tar.gz (83 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m83.5/83.5 kB\u001b[0m \u001b[31m10.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: requests>=2.17.3 in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (2.27.1)\n",
            "Requirement already satisfied: six>=1.10.0 in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (1.16.0)\n",
            "Requirement already satisfied: Flask in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (2.2.5)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (1.22.4)\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (1.5.3)\n",
            "Requirement already satisfied: python-dateutil in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (2.8.2)\n",
            "Requirement already satisfied: protobuf>=3.6.0 in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (3.20.3)\n",
            "Collecting gitpython>=2.1.0 (from mlflow==1.10.0)\n",
            "  Downloading GitPython-3.1.31-py3-none-any.whl (184 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m184.3/184.3 kB\u001b[0m \u001b[31m20.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: pyyaml in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (6.0)\n",
            "Collecting querystring-parser (from mlflow==1.10.0)\n",
            "  Downloading querystring_parser-1.2.4-py2.py3-none-any.whl (7.9 kB)\n",
            "Collecting docker>=4.0.0 (from mlflow==1.10.0)\n",
            "  Downloading docker-6.1.3-py3-none-any.whl (148 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m148.1/148.1 kB\u001b[0m \u001b[31m15.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: entrypoints in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (0.4)\n",
            "Requirement already satisfied: sqlparse in /usr/local/lib/python3.10/dist-packages (from mlflow==1.10.0) (0.4.4)\n",
            "Collecting sqlalchemy<=1.3.13 (from mlflow==1.10.0)\n",
            "  Downloading SQLAlchemy-1.3.13.tar.gz (6.0 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m6.0/6.0 MB\u001b[0m \u001b[31m100.8 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Collecting gorilla (from mlflow==1.10.0)\n",
            "  Downloading gorilla-0.4.0-py2.py3-none-any.whl (9.6 kB)\n",
            "Collecting prometheus-flask-exporter (from mlflow==1.10.0)\n",
            "  Downloading prometheus_flask_exporter-0.22.4-py3-none-any.whl (18 kB)\n",
            "Collecting gunicorn (from mlflow==1.10.0)\n",
            "  Downloading gunicorn-20.1.0-py3-none-any.whl (79 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m79.5/79.5 kB\u001b[0m \u001b[31m9.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting azure-core<2.0.0,>=1.26.0 (from azure-storage-blob>=12.0->mlflow==1.10.0)\n",
            "  Downloading azure_core-1.27.1-py3-none-any.whl (174 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m174.5/174.5 kB\u001b[0m \u001b[31m18.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: cryptography>=2.1.4 in /usr/local/lib/python3.10/dist-packages (from azure-storage-blob>=12.0->mlflow==1.10.0) (41.0.1)\n",
            "Requirement already satisfied: typing-extensions>=4.0.1 in /usr/local/lib/python3.10/dist-packages (from azure-storage-blob>=12.0->mlflow==1.10.0) (4.6.3)\n",
            "Collecting isodate>=0.6.1 (from azure-storage-blob>=12.0->mlflow==1.10.0)\n",
            "  Downloading isodate-0.6.1-py2.py3-none-any.whl (41 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m41.7/41.7 kB\u001b[0m \u001b[31m4.1 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hCollecting pyjwt>=1.7.0 (from databricks-cli>=0.8.7->mlflow==1.10.0)\n",
            "  Downloading PyJWT-2.7.0-py3-none-any.whl (22 kB)\n",
            "Requirement already satisfied: oauthlib>=3.1.0 in /usr/local/lib/python3.10/dist-packages (from databricks-cli>=0.8.7->mlflow==1.10.0) (3.2.2)\n",
            "Requirement already satisfied: tabulate>=0.7.7 in /usr/local/lib/python3.10/dist-packages (from databricks-cli>=0.8.7->mlflow==1.10.0) (0.8.10)\n",
            "Requirement already satisfied: urllib3<2.0.0,>=1.26.7 in /usr/local/lib/python3.10/dist-packages (from databricks-cli>=0.8.7->mlflow==1.10.0) (1.26.16)\n",
            "Requirement already satisfied: packaging>=14.0 in /usr/local/lib/python3.10/dist-packages (from docker>=4.0.0->mlflow==1.10.0) (23.1)\n",
            "Requirement already satisfied: websocket-client>=0.32.0 in /usr/local/lib/python3.10/dist-packages (from docker>=4.0.0->mlflow==1.10.0) (1.6.0)\n",
            "Collecting gitdb<5,>=4.0.1 (from gitpython>=2.1.0->mlflow==1.10.0)\n",
            "  Downloading gitdb-4.0.10-py3-none-any.whl (62 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m62.7/62.7 kB\u001b[0m \u001b[31m7.7 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests>=2.17.3->mlflow==1.10.0) (2023.5.7)\n",
            "Requirement already satisfied: charset-normalizer~=2.0.0 in /usr/local/lib/python3.10/dist-packages (from requests>=2.17.3->mlflow==1.10.0) (2.0.12)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests>=2.17.3->mlflow==1.10.0) (3.4)\n",
            "Collecting Mako (from alembic->mlflow==1.10.0)\n",
            "  Downloading Mako-1.2.4-py3-none-any.whl (78 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m78.7/78.7 kB\u001b[0m \u001b[31m8.9 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: Werkzeug>=2.2.2 in /usr/local/lib/python3.10/dist-packages (from Flask->mlflow==1.10.0) (2.3.6)\n",
            "Requirement already satisfied: Jinja2>=3.0 in /usr/local/lib/python3.10/dist-packages (from Flask->mlflow==1.10.0) (3.1.2)\n",
            "Requirement already satisfied: itsdangerous>=2.0 in /usr/local/lib/python3.10/dist-packages (from Flask->mlflow==1.10.0) (2.1.2)\n",
            "Requirement already satisfied: setuptools>=3.0 in /usr/local/lib/python3.10/dist-packages (from gunicorn->mlflow==1.10.0) (67.7.2)\n",
            "Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.10/dist-packages (from pandas->mlflow==1.10.0) (2022.7.1)\n",
            "Requirement already satisfied: prometheus-client in /usr/local/lib/python3.10/dist-packages (from prometheus-flask-exporter->mlflow==1.10.0) (0.17.0)\n",
            "Requirement already satisfied: cffi>=1.12 in /usr/local/lib/python3.10/dist-packages (from cryptography>=2.1.4->azure-storage-blob>=12.0->mlflow==1.10.0) (1.15.1)\n",
            "Collecting smmap<6,>=3.0.1 (from gitdb<5,>=4.0.1->gitpython>=2.1.0->mlflow==1.10.0)\n",
            "  Downloading smmap-5.0.0-py3-none-any.whl (24 kB)\n",
            "Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.10/dist-packages (from Jinja2>=3.0->Flask->mlflow==1.10.0) (2.1.3)\n",
            "Requirement already satisfied: pycparser in /usr/local/lib/python3.10/dist-packages (from cffi>=1.12->cryptography>=2.1.4->azure-storage-blob>=12.0->mlflow==1.10.0) (2.21)\n",
            "Building wheels for collected packages: databricks-cli, sqlalchemy\n",
            "  Building wheel for databricks-cli (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for databricks-cli: filename=databricks_cli-0.17.7-py3-none-any.whl size=143860 sha256=bc26c8c86f2998bd10e57b98fa28d68f06ad0064e8e261166e12c10c17c5a302\n",
            "  Stored in directory: /root/.cache/pip/wheels/ae/63/93/5402c1a09c1868a59d0b05013484e07af97a9d7b3dbd5bd39a\n",
            "  Building wheel for sqlalchemy (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for sqlalchemy: filename=SQLAlchemy-1.3.13-cp310-cp310-linux_x86_64.whl size=1224536 sha256=284f8979bc665e25a2e829dbb0f0b3b260131801a35ce05cf782169112ed8883\n",
            "  Stored in directory: /root/.cache/pip/wheels/22/e6/98/b450abf168fe8580b4762ab69a7a8efb3dba621866de0dd736\n",
            "Successfully built databricks-cli sqlalchemy\n",
            "Installing collected packages: gorilla, sqlalchemy, smmap, querystring-parser, pyjwt, Mako, isodate, gunicorn, gitdb, docker, databricks-cli, azure-core, alembic, prometheus-flask-exporter, gitpython, azure-storage-blob, mlflow\n",
            "  Attempting uninstall: sqlalchemy\n",
            "    Found existing installation: SQLAlchemy 2.0.16\n",
            "    Uninstalling SQLAlchemy-2.0.16:\n",
            "      Successfully uninstalled SQLAlchemy-2.0.16\n",
            "Successfully installed Mako-1.2.4 alembic-1.11.1 azure-core-1.27.1 azure-storage-blob-12.16.0 databricks-cli-0.17.7 docker-6.1.3 gitdb-4.0.10 gitpython-3.1.31 gorilla-0.4.0 gunicorn-20.1.0 isodate-0.6.1 mlflow-1.10.0 prometheus-flask-exporter-0.22.4 pyjwt-2.7.0 querystring-parser-1.2.4 smmap-5.0.0 sqlalchemy-1.3.13\n",
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Collecting scikit-learn==1.1.0\n",
            "  Downloading scikit_learn-1.1.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (30.7 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m30.7/30.7 MB\u001b[0m \u001b[31m21.6 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: numpy>=1.17.3 in /usr/local/lib/python3.10/dist-packages (from scikit-learn==1.1.0) (1.22.4)\n",
            "Requirement already satisfied: scipy>=1.3.2 in /usr/local/lib/python3.10/dist-packages (from scikit-learn==1.1.0) (1.10.1)\n",
            "Requirement already satisfied: joblib>=1.0.0 in /usr/local/lib/python3.10/dist-packages (from scikit-learn==1.1.0) (1.2.0)\n",
            "Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.10/dist-packages (from scikit-learn==1.1.0) (3.1.0)\n",
            "Installing collected packages: scikit-learn\n",
            "  Attempting uninstall: scikit-learn\n",
            "    Found existing installation: scikit-learn 1.2.2\n",
            "    Uninstalling scikit-learn-1.2.2:\n",
            "      Successfully uninstalled scikit-learn-1.2.2\n",
            "Successfully installed scikit-learn-1.1.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "9AudKP3drpeL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "62ce6db2-fdb8-4e63-9a55-ebe0eebafad7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Numpy: 1.22.4\n",
            "Pandas: 1.5.3\n",
            "matplotlib: 3.7.1\n",
            "seaborn: 0.12.2\n",
            "scikit-learn: 1.1.0\n",
            "MLFlow: 1.10.0\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib #\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "import sklearn #\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.metrics import roc_auc_score, plot_roc_curve, confusion_matrix\n",
        "from sklearn.model_selection import KFold\n",
        "\n",
        "import mlflow\n",
        "import mlflow.sklearn\n",
        "\n",
        "print(\"Numpy: {}\".format(np.__version__))\n",
        "print(\"Pandas: {}\".format(pd.__version__))\n",
        "print(\"matplotlib: {}\".format(matplotlib.__version__))\n",
        "print(\"seaborn: {}\".format(sns.__version__))\n",
        "print(\"scikit-learn: {}\".format(sklearn.__version__))\n",
        "print(\"MLFlow: {}\".format(mlflow.__version__))\n"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Databricks 연결\n",
        "\n",
        "MLFlow Ui 사용을 위해 Databricks에 연결해야 함"
      ],
      "metadata": {
        "id": "2dwM5Xa0Sa5C"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!databricks configure --host https://community.cloud.databricks.com/\n",
        "mlflow.set_tracking_uri(\"databricks\")\n",
        "\n",
        "databricks_email = \"workdofficial@gmail.com\"\n",
        "mlflow_experiment = \"scikit_learn_experiment\"\n",
        "\n",
        "# Databricks에 로그인, experiment가 없을 경우 새로 생성 후 연결\n",
        "mlflow.set_experiment(f\"/Users/{databricks_email}/{mlflow_experiment}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JUeDO8lOSdIi",
        "outputId": "4eea95b6-de8e-4c64-8135-c072e73c5b74"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Username: workdofficial@gmail.com\n",
            "Password: \n",
            "Repeat for confirmation: \n",
            "INFO: '/Users/workdofficial@gmail.com/scikit_learn_experiment' does not exist. Creating a new experiment\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 데이터 처리"
      ],
      "metadata": {
        "id": "0e0mhz8R2-hR"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "아래 사이트에서 신용카드 데이터셋 다운로드\n",
        "- https://www.kaggle.com/datasets/mlg-ulb/creditcardfraud?resource=download\n",
        "\n",
        "다운로드 후 구글 드라이브에 업로드 해두기"
      ],
      "metadata": {
        "id": "btBKwhIazlZV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sNikuq9dYM8U",
        "outputId": "96d6fd43-88d5-4d03-e0fb-fe2afc71510c"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/pexpect/popen_spawn.py:60: DeprecationWarning: setDaemon() is deprecated, set the daemon attribute instead\n",
            "  self._read_thread.setDaemon(True)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "google_drive = \"/content/drive/MyDrive/\"\n",
        "data_path = google_drive + \"creditcard.csv\"\n",
        "df = pd.read_csv(data_path)\n",
        "df = df.drop(\"Time\", axis=1)\n",
        "df.head()"
      ],
      "metadata": {
        "id": "TH_OEwi9tqfb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "outputId": "e1a2b24d-aa19-4806-e734-89472b86862b"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "         V1        V2        V3        V4        V5        V6        V7  \\\n",
              "0 -1.359807 -0.072781  2.536347  1.378155 -0.338321  0.462388  0.239599   \n",
              "1  1.191857  0.266151  0.166480  0.448154  0.060018 -0.082361 -0.078803   \n",
              "2 -1.358354 -1.340163  1.773209  0.379780 -0.503198  1.800499  0.791461   \n",
              "3 -0.966272 -0.185226  1.792993 -0.863291 -0.010309  1.247203  0.237609   \n",
              "4 -1.158233  0.877737  1.548718  0.403034 -0.407193  0.095921  0.592941   \n",
              "\n",
              "         V8        V9       V10  ...       V21       V22       V23       V24  \\\n",
              "0  0.098698  0.363787  0.090794  ... -0.018307  0.277838 -0.110474  0.066928   \n",
              "1  0.085102 -0.255425 -0.166974  ... -0.225775 -0.638672  0.101288 -0.339846   \n",
              "2  0.247676 -1.514654  0.207643  ...  0.247998  0.771679  0.909412 -0.689281   \n",
              "3  0.377436 -1.387024 -0.054952  ... -0.108300  0.005274 -0.190321 -1.175575   \n",
              "4 -0.270533  0.817739  0.753074  ... -0.009431  0.798278 -0.137458  0.141267   \n",
              "\n",
              "        V25       V26       V27       V28  Amount  Class  \n",
              "0  0.128539 -0.189115  0.133558 -0.021053  149.62      0  \n",
              "1  0.167170  0.125895 -0.008983  0.014724    2.69      0  \n",
              "2 -0.327642 -0.139097 -0.055353 -0.059752  378.66      0  \n",
              "3  0.647376 -0.221929  0.062723  0.061458  123.50      0  \n",
              "4 -0.206010  0.502292  0.219422  0.215153   69.99      0  \n",
              "\n",
              "[5 rows x 30 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-6a9f5278-af19-4287-b55e-d8b10c573ad8\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>V1</th>\n",
              "      <th>V2</th>\n",
              "      <th>V3</th>\n",
              "      <th>V4</th>\n",
              "      <th>V5</th>\n",
              "      <th>V6</th>\n",
              "      <th>V7</th>\n",
              "      <th>V8</th>\n",
              "      <th>V9</th>\n",
              "      <th>V10</th>\n",
              "      <th>...</th>\n",
              "      <th>V21</th>\n",
              "      <th>V22</th>\n",
              "      <th>V23</th>\n",
              "      <th>V24</th>\n",
              "      <th>V25</th>\n",
              "      <th>V26</th>\n",
              "      <th>V27</th>\n",
              "      <th>V28</th>\n",
              "      <th>Amount</th>\n",
              "      <th>Class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>-1.359807</td>\n",
              "      <td>-0.072781</td>\n",
              "      <td>2.536347</td>\n",
              "      <td>1.378155</td>\n",
              "      <td>-0.338321</td>\n",
              "      <td>0.462388</td>\n",
              "      <td>0.239599</td>\n",
              "      <td>0.098698</td>\n",
              "      <td>0.363787</td>\n",
              "      <td>0.090794</td>\n",
              "      <td>...</td>\n",
              "      <td>-0.018307</td>\n",
              "      <td>0.277838</td>\n",
              "      <td>-0.110474</td>\n",
              "      <td>0.066928</td>\n",
              "      <td>0.128539</td>\n",
              "      <td>-0.189115</td>\n",
              "      <td>0.133558</td>\n",
              "      <td>-0.021053</td>\n",
              "      <td>149.62</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1.191857</td>\n",
              "      <td>0.266151</td>\n",
              "      <td>0.166480</td>\n",
              "      <td>0.448154</td>\n",
              "      <td>0.060018</td>\n",
              "      <td>-0.082361</td>\n",
              "      <td>-0.078803</td>\n",
              "      <td>0.085102</td>\n",
              "      <td>-0.255425</td>\n",
              "      <td>-0.166974</td>\n",
              "      <td>...</td>\n",
              "      <td>-0.225775</td>\n",
              "      <td>-0.638672</td>\n",
              "      <td>0.101288</td>\n",
              "      <td>-0.339846</td>\n",
              "      <td>0.167170</td>\n",
              "      <td>0.125895</td>\n",
              "      <td>-0.008983</td>\n",
              "      <td>0.014724</td>\n",
              "      <td>2.69</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>-1.358354</td>\n",
              "      <td>-1.340163</td>\n",
              "      <td>1.773209</td>\n",
              "      <td>0.379780</td>\n",
              "      <td>-0.503198</td>\n",
              "      <td>1.800499</td>\n",
              "      <td>0.791461</td>\n",
              "      <td>0.247676</td>\n",
              "      <td>-1.514654</td>\n",
              "      <td>0.207643</td>\n",
              "      <td>...</td>\n",
              "      <td>0.247998</td>\n",
              "      <td>0.771679</td>\n",
              "      <td>0.909412</td>\n",
              "      <td>-0.689281</td>\n",
              "      <td>-0.327642</td>\n",
              "      <td>-0.139097</td>\n",
              "      <td>-0.055353</td>\n",
              "      <td>-0.059752</td>\n",
              "      <td>378.66</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-0.966272</td>\n",
              "      <td>-0.185226</td>\n",
              "      <td>1.792993</td>\n",
              "      <td>-0.863291</td>\n",
              "      <td>-0.010309</td>\n",
              "      <td>1.247203</td>\n",
              "      <td>0.237609</td>\n",
              "      <td>0.377436</td>\n",
              "      <td>-1.387024</td>\n",
              "      <td>-0.054952</td>\n",
              "      <td>...</td>\n",
              "      <td>-0.108300</td>\n",
              "      <td>0.005274</td>\n",
              "      <td>-0.190321</td>\n",
              "      <td>-1.175575</td>\n",
              "      <td>0.647376</td>\n",
              "      <td>-0.221929</td>\n",
              "      <td>0.062723</td>\n",
              "      <td>0.061458</td>\n",
              "      <td>123.50</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-1.158233</td>\n",
              "      <td>0.877737</td>\n",
              "      <td>1.548718</td>\n",
              "      <td>0.403034</td>\n",
              "      <td>-0.407193</td>\n",
              "      <td>0.095921</td>\n",
              "      <td>0.592941</td>\n",
              "      <td>-0.270533</td>\n",
              "      <td>0.817739</td>\n",
              "      <td>0.753074</td>\n",
              "      <td>...</td>\n",
              "      <td>-0.009431</td>\n",
              "      <td>0.798278</td>\n",
              "      <td>-0.137458</td>\n",
              "      <td>0.141267</td>\n",
              "      <td>-0.206010</td>\n",
              "      <td>0.502292</td>\n",
              "      <td>0.219422</td>\n",
              "      <td>0.215153</td>\n",
              "      <td>69.99</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 30 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-6a9f5278-af19-4287-b55e-d8b10c573ad8')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-6a9f5278-af19-4287-b55e-d8b10c573ad8 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-6a9f5278-af19-4287-b55e-d8b10c573ad8');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "normal = df[df.Class == 0].sample(frac=0.5, random_state=2020).reset_index(drop=True)\n",
        "anomaly = df[df.Class == 1]\n",
        "\n",
        "print(f\"Normal: {normal.shape}\")\n",
        "print(f\"Anomaly: {anomaly.shape}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eGjmXf983C2d",
        "outputId": "897a01d0-9d0d-4f96-8043-5539c5549d8b"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Normal: (142158, 30)\n",
            "Anomaly: (492, 30)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# 정상 및 이상 데이터 집합을 학습-테스트-검증 데이터 세트로 분할\n",
        "normal_train, normal_test = train_test_split(normal, test_size = 0.2, random_state = 2020)\n",
        "anomaly_train, anomaly_test = train_test_split(anomaly, test_size = 0.2, random_state = 2020)\n",
        "normal_train, normal_validate = train_test_split(normal_train, test_size = 0.25, random_state = 2020)\n",
        "anomaly_train, anomaly_validate = train_test_split(anomaly_train, test_size = 0.25, random_state = 2020)"
      ],
      "metadata": {
        "id": "EN51c1vJ3TKK"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train = pd.concat((normal_train, anomaly_train))\n",
        "x_test = pd.concat((normal_test, anomaly_test))\n",
        "x_validate = pd.concat((normal_validate, anomaly_validate))\n",
        "\n",
        "y_train = np.array(x_train[\"Class\"])\n",
        "y_test = np.array(x_test[\"Class\"])\n",
        "y_validate = np.array(x_validate[\"Class\"])\n",
        "\n",
        "x_train = x_train.drop(\"Class\", axis=1)\n",
        "x_test = x_test.drop(\"Class\", axis=1)\n",
        "x_validate = x_validate.drop(\"Class\", axis=1)\n",
        "\n",
        "print(\"Training sets:\\nx_train: {} \\ny_train:{}\".format(x_train.shape, y_train.shape))\n",
        "print(\"\\nTesting sets:\\nx_test: {} \\ny_test:{}\".format(x_test.shape, y_test.shape))\n",
        "print(\"\\nValidate sets:\\nx_validate: {} \\ny_validate:{}\".format(x_validate.shape, y_validate.shape))\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YSnwAtFj4GOx",
        "outputId": "3b50db11-a0d1-4ea1-d4d7-6b708c71eeba"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Training sets:\n",
            "x_train: (85588, 29) \n",
            "y_train:(85588,)\n",
            "\n",
            "Testing sets:\n",
            "x_test: (28531, 29) \n",
            "y_test:(28531,)\n",
            "\n",
            "Validate sets:\n",
            "x_validate: (28531, 29) \n",
            "y_validate:(28531,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### 스케일러를 통한 데이터 확장"
      ],
      "metadata": {
        "id": "T_9v67A04YNU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "scaler = StandardScaler()\n",
        "scaler.fit(pd.concat((normal, anomaly)).drop(\"Class\", axis=1))\n",
        "x_train = scaler.transform(x_train)\n",
        "x_test = scaler.transform(x_test)\n",
        "x_validate = scaler.transform(x_validate)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8bGU0fkw6WBK",
        "outputId": "9182c9f0-e4d9-402f-cc5a-0e7ca34dd7cd"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.\n",
            "  and should_run_async(code)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# MLFlow를 통한 학습 및 평가 (p130)"
      ],
      "metadata": {
        "id": "aRhHnY204n-a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def train(sk_model, x_train, y_train):\n",
        "  sk_model = sk_model.fit(x_train, y_train)\n",
        "\n",
        "  train_acc = sk_model.score(x_train, y_train)\n",
        "\n",
        "  # MLFlow에 학습 정확도값을 추적하도록 설정\n",
        "  mlflow.log_metric(\"train_acc\", train_acc)\n",
        "\n",
        "  print(f\"Train Accuracy: {train_acc:.3%}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rspFossV4pqz",
        "outputId": "abfc4ec7-d9cf-4867-b75c-da1fa9e0fd53"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.\n",
            "  and should_run_async(code)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def evaluate(sk_model, x_test, y_test):\n",
        "  eval_acc = sk_model.score(x_test, y_test)\n",
        "\n",
        "  preds = sk_model.predict(x_test)\n",
        "  auc_score = roc_auc_score(y_test, preds)\n",
        "\n",
        "  # MLFlow에 테스트 정확도 및 정확도 점수를 추적하도록 설정\n",
        "  mlflow.log_metric(\"eval_acc\", eval_acc)\n",
        "  mlflow.log_metric(\"auc_score\", auc_score)\n",
        "\n",
        "  print(f\"Auc Score: {auc_score:.3%}\")\n",
        "  print(f\"Eval Accuracy: {eval_acc:.3%}\")\n",
        "\n",
        "  roc_plot = plot_roc_curve(sk_model, x_test, y_test, name ='scikit-learn ROC Curve')\n",
        "  plt.savefig(\"sklearn_roc_plot.png\")\n",
        "  plt.show()\n",
        "  plt.clf()\n",
        "\n",
        "  conf_matrix = confusion_matrix(y_test, preds)\n",
        "  ax = sns.heatmap(conf_matrix, annot=True, fmt='g')\n",
        "  ax.invert_xaxis()\n",
        "  ax.invert_yaxis()\n",
        "  plt.ylabel('Actual')\n",
        "  plt.xlabel('Predicted')\n",
        "  plt.title(\"Confusion Matrix\")\n",
        "  plt.savefig(\"sklearn_conf_matrix.png\")\n",
        "\n",
        "  # MLFlow에 로컬에 저장된 plot을 로깅하도록 설정\n",
        "  mlflow.log_artifact(\"sklearn_roc_plot.png\")\n",
        "  mlflow.log_artifact(\"sklearn_conf_matrix.png\")"
      ],
      "metadata": {
        "id": "v58msZRd6P7k"
      },
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# MLFlow 실행 로깅 및 확인 (p133)"
      ],
      "metadata": {
        "id": "iaKYQTV-BEBH"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "sk_model = LogisticRegression(random_state=None, max_iter=400, solver='newton-cg')\n",
        "\n",
        "# MLFlow 실험 이름 설정\n",
        "mlflow.set_experiment(f\"/Users/{databricks_email}/scikit_learn_experiment\")\n",
        "\n",
        "# MLFlow 실험 시작\n",
        "with mlflow.start_run():\n",
        "  train(sk_model, x_train, y_train)\n",
        "  evaluate(sk_model, x_test, y_test)\n",
        "\n",
        "  # MLFlow에 sk_model 모델 로깅\n",
        "  mlflow.sklearn.log_model(sk_model, \"log_reg_model\")\n",
        "\n",
        "  # 로깅된 데이터를 출력\n",
        "  print(\"Model run: \", mlflow.active_run().info.run_uuid)\n",
        "\n",
        "mlflow.end_run()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "N7u_8t1d8i3k",
        "outputId": "0f9ad255-35d3-4c22-dc05-41875e3de55f"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Train Accuracy: 99.893%\n",
            "Auc Score: 85.341%\n",
            "Eval Accuracy: 99.874%\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function plot_roc_curve is deprecated; Function :func:`plot_roc_curve` is deprecated in 1.0 and will be removed in 1.2. Use one of the class methods: :meth:`sklearn.metric.RocCurveDisplay.from_predictions` or :meth:`sklearn.metric.RocCurveDisplay.from_estimator`.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model run:  af6dc2f1ce3d46d38bd5da8481afc358\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# MLFlow를 사용한 모델 검증 (파라미터 튜닝) (p143)"
      ],
      "metadata": {
        "id": "tkPzUK-rBOM9"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "anomaly_weights = [1, 5, 10, 15]\n",
        "num_folds = 5\n",
        "kfold = KFold(n_splits=num_folds, shuffle=True, random_state=2020)"
      ],
      "metadata": {
        "id": "B3qe4vJ_A37q"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "mlflow.set_experiment(f\"/Users/{databricks_email}/sklearn_creditcard_broad_search\")\n",
        "\n",
        "logs = []\n",
        "\n",
        "for f in range(len(anomaly_weights)):\n",
        "  fold = 1\n",
        "  accuracies = []\n",
        "  auc_scores = []\n",
        "\n",
        "  for train, test in kfold.split(x_validate, y_validate):\n",
        "    with mlflow.start_run():\n",
        "      weight = anomaly_weights[f]\n",
        "      mlflow.log_param(\"anomaly_weight\", weight)\n",
        "      class_weights = {\n",
        "          0: 1,\n",
        "          1: weight\n",
        "      }\n",
        "\n",
        "      sk_model = LogisticRegression(random_state=None,\n",
        "                                    max_iter=400,\n",
        "                                    solver='newton-cg',\n",
        "                                    class_weight=class_weights).fit(x_validate[train], y_validate[train])\n",
        "\n",
        "      for h in range(40):\n",
        "        print('-', end=\"\")\n",
        "\n",
        "      print(f\"\\nfold {fold}\\nAnomaly Weight: {weight}\")\n",
        "\n",
        "      train_acc = sk_model.score(x_validate[train], y_validate[train])\n",
        "\n",
        "      mlflow.log_metric(\"train_acc\", train_acc)\n",
        "\n",
        "      eval_acc = sk_model.score(x_validate[test], y_validate[test])\n",
        "      preds = sk_model.predict(x_validate[test])\n",
        "\n",
        "      mlflow.log_metric(\"eval_acc\", eval_acc)\n",
        "\n",
        "      try:\n",
        "        auc_score = roc_auc_score(y_validate[test], preds)\n",
        "\n",
        "      except:\n",
        "        auc_score = -1\n",
        "\n",
        "      mlflow.log_metric(\"auc_score\", auc_score)\n",
        "\n",
        "      print(\"AUC: {}\\neval_acc: {}\".format(auc_score, eval_acc))\n",
        "\n",
        "      accuracies.append(eval_acc)\n",
        "      auc_scores.append(auc_score)\n",
        "\n",
        "      log = [sk_model, x_validate[test], y_validate[test], preds]\n",
        "      logs.append(log)\n",
        "\n",
        "      mlflow.sklearn.log_model(sk_model, f\"anom_weight_{weight}_fold_{fold}\")\n",
        "\n",
        "      fold = fold + 1\n",
        "\n",
        "      mlflow.end_run()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FeZ6uDbBBLrj",
        "outputId": "6721be46-9fce-42f6-8fb4-b3c8a03a8a67"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "INFO: '/Users/workdofficial@gmail.com/sklearn_creditcard_broad_search' does not exist. Creating a new experiment\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 1\n",
            "AUC: 0.8\n",
            "eval_acc: 0.9985982127212195\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 1\n",
            "AUC: 0.8568790049001132\n",
            "eval_acc: 0.998422712933754\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 1\n",
            "AUC: 0.8406451922461774\n",
            "eval_acc: 0.9982474588152822\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 1\n",
            "AUC: 0.8682452129973255\n",
            "eval_acc: 0.9987732211706976\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 1\n",
            "AUC: 0.8528532875621686\n",
            "eval_acc: 0.9989484752891693\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 5\n",
            "AUC: 0.8748241603657465\n",
            "eval_acc: 0.9987734361310671\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 5\n",
            "AUC: 0.975398919462244\n",
            "eval_acc: 0.9982474588152822\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 5\n",
            "AUC: 0.9314663169342972\n",
            "eval_acc: 0.9987732211706976\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 5\n",
            "AUC: 0.8943851628367561\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 5\n",
            "AUC: 0.8820013855427915\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.9243845612801126\n",
            "eval_acc: 0.9982477659015244\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.9753109687146626\n",
            "eval_acc: 0.9980722046968104\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.9313783507133262\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.8943851628367561\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.8820013855427915\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 15\n",
            "AUC: 0.9242966414629858\n",
            "eval_acc: 0.9980725424916769\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 15\n",
            "AUC: 0.9751350672194998\n",
            "eval_acc: 0.9977216964598669\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 15\n",
            "AUC: 0.9312903844923551\n",
            "eval_acc: 0.998422712933754\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 15\n",
            "AUC: 0.8940334835682489\n",
            "eval_acc: 0.9978969505783386\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 15\n",
            "AUC: 0.8813861631838532\n",
            "eval_acc: 0.9973711882229233\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"\\nAverages: \")\n",
        "print(\"Accuracy: \", np.mean(accuracies))\n",
        "print(\"AUC: \", np.mean(auc_scores))\n",
        "print(\"Best: \")\n",
        "print(\"Accuracy: \", np.max(accuracies))\n",
        "print(\"AUC: \", np.max(auc_scores))\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2ejJnAFkH5ix",
        "outputId": "58247ff5-4828-49e4-cc4e-b2fafac0501c"
      },
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Averages: \n",
            "Accuracy:  0.9978970181373118\n",
            "AUC:  0.9212283479853886\n",
            "Best: \n",
            "Accuracy:  0.998422712933754\n",
            "AUC:  0.9751350672194998\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.\n",
            "  and should_run_async(code)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 파라미터 튜닝 - Guided Search (p156)"
      ],
      "metadata": {
        "id": "rwhQZ5QxWTfI"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "anomaly_weights = [10, 50, 100, 150, 200]\n",
        "num_folds = 5\n",
        "kfold = KFold(n_splits=num_folds, shuffle=True, random_state=2020)"
      ],
      "metadata": {
        "id": "JTa2LfmIWjCu"
      },
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "mlflow.set_experiment(f\"/Users/{databricks_email}/sklearn_creditcard_guided_search\")\n",
        "\n",
        "logs = []\n",
        "\n",
        "for f in range(len(anomaly_weights)):\n",
        "  fold = 1\n",
        "  accuracies = []\n",
        "  auc_scores = []\n",
        "\n",
        "  for train, test in kfold.split(x_validate, y_validate):\n",
        "    with mlflow.start_run():\n",
        "      weight = anomaly_weights[f]\n",
        "      mlflow.log_param(\"anomaly_weight\", weight)\n",
        "      class_weights = {\n",
        "          0: 1,\n",
        "          1: weight\n",
        "      }\n",
        "\n",
        "      sk_model = LogisticRegression(random_state=None,\n",
        "                                    max_iter=400,\n",
        "                                    solver='newton-cg',\n",
        "                                    class_weight=class_weights).fit(x_validate[train], y_validate[train])\n",
        "\n",
        "      for h in range(40):\n",
        "        print('-', end=\"\")\n",
        "\n",
        "      print(f\"\\nfold {fold}\\nAnomaly Weight: {weight}\")\n",
        "\n",
        "      train_acc = sk_model.score(x_validate[train], y_validate[train])\n",
        "\n",
        "      mlflow.log_metric(\"train_acc\", train_acc)\n",
        "\n",
        "      eval_acc = sk_model.score(x_validate[test], y_validate[test])\n",
        "      preds = sk_model.predict(x_validate[test])\n",
        "\n",
        "      mlflow.log_metric(\"eval_acc\", eval_acc)\n",
        "\n",
        "      try:\n",
        "        auc_score = roc_auc_score(y_validate[test], preds)\n",
        "\n",
        "      except:\n",
        "        auc_score = -1\n",
        "\n",
        "      mlflow.log_metric(\"auc_score\", auc_score)\n",
        "\n",
        "      print(\"AUC: {}\\neval_acc: {}\".format(auc_score, eval_acc))\n",
        "\n",
        "      accuracies.append(eval_acc)\n",
        "      auc_scores.append(auc_score)\n",
        "\n",
        "      log = [sk_model, x_validate[test], y_validate[test], preds]\n",
        "      logs.append(log)\n",
        "\n",
        "      mlflow.sklearn.log_model(sk_model, f\"anom_weight_{weight}_fold_{fold}\")\n",
        "\n",
        "      fold = fold + 1\n",
        "\n",
        "      mlflow.end_run()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IGUXd_YEWvzP",
        "outputId": "089c7236-92d6-4a2d-c559-f35d8fd63e8f"
      },
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "INFO: '/Users/workdofficial@gmail.com/sklearn_creditcard_guided_search' does not exist. Creating a new experiment\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.9243845612801126\n",
            "eval_acc: 0.9982477659015244\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.9753109687146626\n",
            "eval_acc: 0.9980722046968104\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.9313783507133262\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.8943851628367561\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 10\n",
            "AUC: 0.8820013855427915\n",
            "eval_acc: 0.9985979670522257\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 50\n",
            "AUC: 0.922626164937577\n",
            "eval_acc: 0.9947432977045734\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 50\n",
            "AUC: 0.9734640030154541\n",
            "eval_acc: 0.9943918682089029\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 50\n",
            "AUC: 0.9296190262939033\n",
            "eval_acc: 0.99509288468279\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 50\n",
            "AUC: 0.9183271172480171\n",
            "eval_acc: 0.9940413599719593\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 50\n",
            "AUC: 0.8797162739238779\n",
            "eval_acc: 0.9940413599719593\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 100\n",
            "AUC: 0.9205160893265342\n",
            "eval_acc: 0.990537935868232\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 100\n",
            "AUC: 0.9951627088830255\n",
            "eval_acc: 0.9903610234840519\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 100\n",
            "AUC: 0.9268920734437976\n",
            "eval_acc: 0.9896600070101648\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 100\n",
            "AUC: 0.9429724301962926\n",
            "eval_acc: 0.9908867858394672\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 100\n",
            "AUC: 0.8777827179386432\n",
            "eval_acc: 0.9901857693655801\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 150\n",
            "AUC: 0.9186697731668719\n",
            "eval_acc: 0.9868582442614333\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 150\n",
            "AUC: 0.9927000879507476\n",
            "eval_acc: 0.985453908166842\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 150\n",
            "AUC: 0.9247808841404901\n",
            "eval_acc: 0.985453908166842\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 150\n",
            "AUC: 0.9411261140366303\n",
            "eval_acc: 0.9872064493515598\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 150\n",
            "AUC: 0.9053488155677106\n",
            "eval_acc: 0.9866806869961444\n",
            "----------------------------------------\n",
            "fold 1\n",
            "Anomaly Weight: 200\n",
            "AUC: 0.9427026551784772\n",
            "eval_acc: 0.9851060101629577\n",
            "----------------------------------------\n",
            "fold 2\n",
            "Anomaly Weight: 200\n",
            "AUC: 0.9904133685136324\n",
            "eval_acc: 0.9808973010865756\n",
            "----------------------------------------\n",
            "fold 3\n",
            "Anomaly Weight: 200\n",
            "AUC: 0.9226696948371826\n",
            "eval_acc: 0.9812478093235191\n",
            "----------------------------------------\n",
            "fold 4\n",
            "Anomaly Weight: 200\n",
            "AUC: 0.9391039582427143\n",
            "eval_acc: 0.9831756046267087\n",
            "----------------------------------------\n",
            "fold 5\n",
            "Anomaly Weight: 200\n",
            "AUC: 0.9029758150403772\n",
            "eval_acc: 0.9819488257974063\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"\\nAverages: \")\n",
        "print(\"Accuracy: \", np.mean(accuracies))\n",
        "print(\"AUC: \", np.mean(auc_scores))\n",
        "print(\"Best: \")\n",
        "print(\"Accuracy: \", np.max(accuracies))\n",
        "print(\"AUC: \", np.max(auc_scores))\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ak1K7siRY7C8",
        "outputId": "aa21dafa-e09e-42df-a1eb-8b4f54e88671"
      },
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Averages: \n",
            "Accuracy:  0.9824751101994336\n",
            "AUC:  0.9395730983624768\n",
            "Best: \n",
            "Accuracy:  0.9851060101629577\n",
            "AUC:  0.9904133685136324\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "DEq8PbFMZBap"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}