OnlineRetail_clustering.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "OnlineRetail_clustering.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/imji0319/PDSH/blob/master/OnlineRetail_clustering.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "ZBaEtMlWF6gN",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# RFM 기반 고객 세그멘테이션"
      ]
    },
    {
      "metadata": {
        "id": "Ygv3kV5i-17H",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "c6a46eba-cb66-417f-94d7-b75c5962b4bd"
      },
      "cell_type": "code",
      "source": [
        "from google.colab import auth\n",
        "auth.authenticate_user()\n",
        "\n",
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "rEaPt-zA_biG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "8beb0f4e-8b82-4a42-b0bc-badeac036d48"
      },
      "cell_type": "code",
      "source": [
        "cd /content/gdrive/My Drive/data"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/gdrive/My Drive/data\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "UGzkWb9y-7Jy",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 142
        },
        "outputId": "461b01e4-e223-4dfc-ac08-7140dfc56808"
      },
      "cell_type": "code",
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import datetime\n",
        "import math\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline \n",
        "\n",
        "\n",
        "retail_df = pd.read_excel(io='Online Retail.xlsx')\n",
        "retail_df.head(3)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>InvoiceNo</th>\n",
              "      <th>StockCode</th>\n",
              "      <th>Description</th>\n",
              "      <th>Quantity</th>\n",
              "      <th>InvoiceDate</th>\n",
              "      <th>UnitPrice</th>\n",
              "      <th>CustomerID</th>\n",
              "      <th>Country</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>536365</td>\n",
              "      <td>85123A</td>\n",
              "      <td>WHITE HANGING HEART T-LIGHT HOLDER</td>\n",
              "      <td>6</td>\n",
              "      <td>2010-12-01 08:26:00</td>\n",
              "      <td>2.55</td>\n",
              "      <td>17850.0</td>\n",
              "      <td>United Kingdom</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>536365</td>\n",
              "      <td>71053</td>\n",
              "      <td>WHITE METAL LANTERN</td>\n",
              "      <td>6</td>\n",
              "      <td>2010-12-01 08:26:00</td>\n",
              "      <td>3.39</td>\n",
              "      <td>17850.0</td>\n",
              "      <td>United Kingdom</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>536365</td>\n",
              "      <td>84406B</td>\n",
              "      <td>CREAM CUPID HEARTS COAT HANGER</td>\n",
              "      <td>8</td>\n",
              "      <td>2010-12-01 08:26:00</td>\n",
              "      <td>2.75</td>\n",
              "      <td>17850.0</td>\n",
              "      <td>United Kingdom</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "  InvoiceNo StockCode                         Description  Quantity  \\\n",
              "0    536365    85123A  WHITE HANGING HEART T-LIGHT HOLDER         6   \n",
              "1    536365     71053                 WHITE METAL LANTERN         6   \n",
              "2    536365    84406B      CREAM CUPID HEARTS COAT HANGER         8   \n",
              "\n",
              "          InvoiceDate  UnitPrice  CustomerID         Country  \n",
              "0 2010-12-01 08:26:00       2.55     17850.0  United Kingdom  \n",
              "1 2010-12-01 08:26:00       3.39     17850.0  United Kingdom  \n",
              "2 2010-12-01 08:26:00       2.75     17850.0  United Kingdom  "
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "metadata": {
        "id": "Adozj9rx_S7k",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "33658b62-e4a0-47a1-f785-29804b7948b7"
      },
      "cell_type": "code",
      "source": [
        "retail_df.shape"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(541909, 8)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "metadata": {
        "id": "nH8Fly9N_5Xu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 238
        },
        "outputId": "114584b5-20e4-4318-fb16-3606b634b9c7"
      },
      "cell_type": "code",
      "source": [
        "retail_df.info()"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 541909 entries, 0 to 541908\n",
            "Data columns (total 8 columns):\n",
            "InvoiceNo      541909 non-null object\n",
            "StockCode      541909 non-null object\n",
            "Description    540455 non-null object\n",
            "Quantity       541909 non-null int64\n",
            "InvoiceDate    541909 non-null datetime64[ns]\n",
            "UnitPrice      541909 non-null float64\n",
            "CustomerID     406829 non-null float64\n",
            "Country        541909 non-null object\n",
            "dtypes: datetime64[ns](1), float64(2), int64(1), object(4)\n",
            "memory usage: 33.1+ MB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "v1mY_VVOATYK",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- NULL 데이터 제거 : customer segmentation을 위해 customerID가 없는 경우 삭제 \n",
        "- 오류 데이터 제거\n",
        "    1. Quantity < 0 \n",
        "    2. UnitPrice < 0 "
      ]
    },
    {
      "metadata": {
        "id": "pA1zAKFjABvu",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        },
        "outputId": "44c60b05-d0d3-4612-e780-f49f3b69084f"
      },
      "cell_type": "code",
      "source": [
        "retail_df = retail_df[retail_df['Quantity'] > 0]\n",
        "retail_df = retail_df[retail_df['UnitPrice'] > 0]\n",
        "retail_df = retail_df[retail_df['CustomerID'].notnull()]\n",
        "\n",
        "print(retail_df.shape)\n",
        "retail_df.isnull().sum()"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(397884, 8)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "InvoiceNo      0\n",
              "StockCode      0\n",
              "Description    0\n",
              "Quantity       0\n",
              "InvoiceDate    0\n",
              "UnitPrice      0\n",
              "CustomerID     0\n",
              "Country        0\n",
              "dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "metadata": {
        "id": "2NBsj80gBIhf",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "- 주요 주문 고객이 영국으로 그 외 다른 국가의 데이터는 모두 삭제"
      ]
    },
    {
      "metadata": {
        "id": "wqk7NSFnBDBK",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "0d56482d-b379-4fd4-d4ce-47753df77b42"
      },
      "cell_type": "code",
      "source": [
        "retail_df = retail_df[retail_df['Country'] == 'United Kingdom']\n",
        "print(retail_df.shape)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(354321, 8)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "m1Yz4hyrBgMw",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "필요한 데이터 가공\n",
        "- 주문 금액 데이터 (sale_amount) = 주문수량 (Quantity)  * 제품 단가(UnitPrice)\n",
        "- CustomerID 형태 int 로 변환 "
      ]
    },
    {
      "metadata": {
        "id": "Ynu8UycOBctx",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "retail_df['sale_amount'] = retail_df['Quantity'] * retail_df['UnitPrice']\n",
        "retail_df['CustomerID'] = retail_df['CustomerID'].astype(int)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "nlq6oakrCRHp",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 238
        },
        "outputId": "eb4f18e6-ec96-4d78-f54f-3d8de4773ccd"
      },
      "cell_type": "code",
      "source": [
        "# TOP-5 주문 건수와 주문 금액을 가진 고객 데이터를 추출\n",
        "print( retail_df['CustomerID'].value_counts().head(5) )\n",
        "print( retail_df.groupby('CustomerID')['sale_amount'].sum().sort_values(ascending=False)[:5])"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "17841    7847\n",
            "14096    5111\n",
            "12748    4595\n",
            "14606    2700\n",
            "15311    2379\n",
            "Name: CustomerID, dtype: int64\n",
            "CustomerID\n",
            "18102    259657.30\n",
            "17450    194550.79\n",
            "16446    168472.50\n",
            "17511     91062.38\n",
            "16029     81024.84\n",
            "Name: sale_amount, dtype: float64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "OEw_plNOCzZO",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "78cc8985-af08-4ee3-a226-e9563c54e633"
      },
      "cell_type": "code",
      "source": [
        "retail_df.groupby(['InvoiceNo', 'StockCode'])['InvoiceNo'].count().mean()"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1.028702077315023"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "metadata": {
        "id": "wup_RmKFEIbg",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "DateFrame 의 groupby() 의 multiple 연산을 위해 agg() 이용\n",
        "- Recency : InvoiceDate의 max() \n",
        "- Frequency :  InvoiceNo의 count()\n",
        "- Monetary : sale_amount의 sum()"
      ]
    },
    {
      "metadata": {
        "id": "xhmUB_tlDY-2",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 142
        },
        "outputId": "341086b3-5f46-4ba5-83a8-c83cab3f51c7"
      },
      "cell_type": "code",
      "source": [
        "aggregations = {\n",
        "    \n",
        "    'InvoiceDate' : 'max',\n",
        "    'InvoiceNo':'count',\n",
        "    'sale_amount':'sum'\n",
        "    \n",
        "}\n",
        "\n",
        "cust_df = retail_df.groupby('CustomerID').agg(aggregations)\n",
        "cust_df = cust_df.rename(columns = {'InvoiceDate' : 'Recency',\n",
        "                                                            'InvoiceNo' : 'Frequency',\n",
        "                                                            'sale_amount' : 'Monetary'}\n",
        "                                        )\n",
        "\n",
        "cust_df = cust_df.reset_index()\n",
        "cust_df.head(3)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>CustomerID</th>\n",
              "      <th>Recency</th>\n",
              "      <th>Frequency</th>\n",
              "      <th>Monetary</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>12346</td>\n",
              "      <td>2011-01-18 10:01:00</td>\n",
              "      <td>1</td>\n",
              "      <td>77183.60</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>12747</td>\n",
              "      <td>2011-12-07 14:34:00</td>\n",
              "      <td>103</td>\n",
              "      <td>4196.01</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>12748</td>\n",
              "      <td>2011-12-09 12:20:00</td>\n",
              "      <td>4595</td>\n",
              "      <td>33719.73</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   CustomerID             Recency  Frequency  Monetary\n",
              "0       12346 2011-01-18 10:01:00          1  77183.60\n",
              "1       12747 2011-12-07 14:34:00        103   4196.01\n",
              "2       12748 2011-12-09 12:20:00       4595  33719.73"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "metadata": {
        "id": "Nekj3V3mFJkR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 159
        },
        "outputId": "ea131f68-b255-4f4a-f33a-023456c68333"
      },
      "cell_type": "code",
      "source": [
        "import datetime as dt\n",
        "\n",
        "cust_df['Recency'] = dt.datetime(2011,12,10) - cust_df['Recency']\n",
        "cust_df['Recency'] = cust_df['Recency'].apply(lambda x: x.days+1)\n",
        "\n",
        "print('cust_df 로우와 칼럼 건수는 ', cust_df.shape)\n",
        "cust_df.head(3)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "cust_df 로우와 칼럼 건수는  (3920, 4)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>CustomerID</th>\n",
              "      <th>Recency</th>\n",
              "      <th>Frequency</th>\n",
              "      <th>Monetary</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>12346</td>\n",
              "      <td>326</td>\n",
              "      <td>1</td>\n",
              "      <td>77183.60</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>12747</td>\n",
              "      <td>3</td>\n",
              "      <td>103</td>\n",
              "      <td>4196.01</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>12748</td>\n",
              "      <td>1</td>\n",
              "      <td>4595</td>\n",
              "      <td>33719.73</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   CustomerID  Recency  Frequency  Monetary\n",
              "0       12346      326          1  77183.60\n",
              "1       12747        3        103   4196.01\n",
              "2       12748        1       4595  33719.73"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "metadata": {
        "id": "NkAefsAtGJ3W",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "온라인 판매 데이터 세틑는 소매업체의 대규모 주문을 포함\n",
        "- 주문 횟수와 주문 금액에서 개인 고객 주문과 매우 큰 차이를 보임 \n",
        "- 왜곡된 데이터 분포도를 가지게 되어 군집화가 한쪽 군집에만 집중되는 현상 발생"
      ]
    },
    {
      "metadata": {
        "id": "MZLKoPZjFr-W",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 383
        },
        "outputId": "6b9d0df9-2135-4ab2-eb23-bfa5d0f60d42"
      },
      "cell_type": "code",
      "source": [
        "fig, (ax1, ax2, ax3) =plt.subplots(figsize=(12,4), nrows =1 , ncols = 3)\n",
        "\n",
        "ax1.set_title('Recency Histogram')\n",
        "ax1.hist(cust_df['Recency'])\n",
        "\n",
        "\n",
        "ax2.set_title('Frequecy Histogram')\n",
        "ax2.hist(cust_df['Frequency'])\n",
        "\n",
        "ax3.set_title('Monetary Histogram')\n",
        "ax3.hist(cust_df['Monetary'])"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(array([3.887e+03, 1.900e+01, 9.000e+00, 2.000e+00, 0.000e+00, 0.000e+00,\n",
              "        1.000e+00, 1.000e+00, 0.000e+00, 1.000e+00]),\n",
              " array([3.75000000e+00, 2.59691050e+04, 5.19344600e+04, 7.78998150e+04,\n",
              "        1.03865170e+05, 1.29830525e+05, 1.55795880e+05, 1.81761235e+05,\n",
              "        2.07726590e+05, 2.33691945e+05, 2.59657300e+05]),\n",
              " <a list of 10 Patch objects>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAs8AAAEICAYAAACgdxkmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMfPJNxZGAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 864x288 with 3 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "5RlqMG-FG9l9",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "outputId": "781f20ee-d36d-4db1-d00c-9ffe9568e87d"
      },
      "cell_type": "code",
      "source": [
        "# 데이터 값의 백분위 \n",
        "\n",
        "cust_df[['Recency' ,'Frequency','Monetary']].describe()"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>Recency</th>\n",
              "      <th>Frequency</th>\n",
              "      <th>Monetary</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>3920.000000</td>\n",
              "      <td>3920.000000</td>\n",
              "      <td>3920.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>92.742092</td>\n",
              "      <td>90.388010</td>\n",
              "      <td>1864.385601</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>99.533485</td>\n",
              "      <td>217.808385</td>\n",
              "      <td>7482.817477</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>1.000000</td>\n",
              "      <td>1.000000</td>\n",
              "      <td>3.750000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>18.000000</td>\n",
              "      <td>17.000000</td>\n",
              "      <td>300.280000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>51.000000</td>\n",
              "      <td>41.000000</td>\n",
              "      <td>652.280000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>143.000000</td>\n",
              "      <td>99.250000</td>\n",
              "      <td>1576.585000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>374.000000</td>\n",
              "      <td>7847.000000</td>\n",
              "      <td>259657.300000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "           Recency    Frequency       Monetary\n",
              "count  3920.000000  3920.000000    3920.000000\n",
              "mean     92.742092    90.388010    1864.385601\n",
              "std      99.533485   217.808385    7482.817477\n",
              "min       1.000000     1.000000       3.750000\n",
              "25%      18.000000    17.000000     300.280000\n",
              "50%      51.000000    41.000000     652.280000\n",
              "75%     143.000000    99.250000    1576.585000\n",
              "max     374.000000  7847.000000  259657.300000"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 23
        }
      ]
    },
    {
      "metadata": {
        "id": "k4Ho9-f8HXSk",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "왜곡이 심한 데이터 그대로 Kmeans 군집화를 할 경우 변별력이 떨어짐 -> StandardScaler"
      ]
    },
    {
      "metadata": {
        "id": "98mF3Vq9HI8x",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "f4f40312-5584-461c-faa3-15eacb3292c3"
      },
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.cluster import KMeans\n",
        "from sklearn.metrics import silhouette_samples, silhouette_score\n",
        "\n",
        "\n",
        "X_features = cust_df[['Recency', 'Frequency', 'Monetary' ]].values\n",
        "X_features_scaled = StandardScaler().fit_transform(X_features)\n",
        "\n",
        "kmeans = KMeans(n_clusters = 3, random_state = 0)\n",
        "labels = kmeans.fit_predict(X_features_scaled)\n",
        "cust_df['cluster_label'] = labels \n",
        "\n",
        "print('실루엣 스코어는 : {0:.3f}'.format(silhouette_score(X_features_scaled, labels)))"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "실루엣 스코어는 : 0.592\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "Z9Mrdt2JMVWf",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "ecfa4bf5-3316-495e-fb6e-4cb3eeb6a82b"
      },
      "cell_type": "code",
      "source": [
        "# 로그 변환 \n",
        "\n",
        "cust_df['Recency_log'] = np.log1p(cust_df['Recency'])\n",
        "cust_df['Frequency_log'] = np.log1p(cust_df['Frequency'])\n",
        "cust_df['Monetary_log'] = np.log1p(cust_df['Monetary'])\n",
        "\n",
        "\n",
        "X_features = cust_df[['Recency_log', 'Frequency_log', 'Monetary_log' ]].values\n",
        "X_features_scaled = StandardScaler().fit_transform(X_features)\n",
        "\n",
        "kmeans = KMeans(n_clusters = 3, random_state = 0)\n",
        "labels = kmeans.fit_predict(X_features_scaled)\n",
        "cust_df['cluster_label'] = labels \n",
        "\n",
        "\n",
        "print('실루엣 스코어는 : {0:.3f}'.format(silhouette_score(X_features_scaled, labels)))\n"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "실루엣 스코어는 : 0.305\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}