Pt2.Modeling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# <center><span style=\"color:DarkBlue\">Modeling</span></center> <a class=\"tocSkip\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 변수 설명\n",
    "\\**일차적으로 다음의 변수들은 모델링에 사용하지 않기로 함.*\n",
    "\n",
    "- ~LoanNr_ChkDgt : 식별 키~\n",
    "- ~Name : 차용 기업명(Borrower)~\n",
    "- ~City : 차용 기업이 위치한 도시~\n",
    "- ~State : 차용 기업이 위치한 주~\n",
    "- ~Zip : 차용 기업의 우편 번호~\n",
    "- ~Bank : 은행 이름~\n",
    "- ~BankState : 은행이 위치한 주~\n",
    "- NAICS : 북미 산업 분류 체계 코드\n",
    "- ~ApprovalDate : SBA 약정 발행 날짜~\n",
    "- ~ApprovalFY : 약정 회계 연도~\n",
    "- ~Term : 대출 기간(개월)~\n",
    "- NoEmp : 직원 수\n",
    "- NewExist : 0 = 기존 비즈니스, 1 = 신규 비즈니스\n",
    "- ~CreateJob : 신규 직원 수~\n",
    "- ~RetainedJob : 기존 직원 수~\n",
    "- FranchiseCode\t: 0=프랜차이즈 아님, 1=프랜차이즈\n",
    "- UrbanRural : 0 = 도시, 1 = 농촌\n",
    "- LowDoc : SBA 대출프로그램 신청 여부(0=아니오, 1 = 예)\n",
    "- ~ChgOffDate : 채무 불이행으로 선언되는 날짜~\n",
    "- ~DisbursementDate : 대금 지급일~\n",
    "- ~DisbursementGross : 지급된 금액~\n",
    "- `MIS_Status` : 채무불이행 여부 (0=채무이행, 1=채무불이행)\n",
    "- ~ChgOffPrinGr : 청구된 금액~\n",
    "- GrAppv : 은행에서 승인한 대출 금액\n",
    "- SBA_Appv : SBA의 승인된 대출 보증 금액\n",
    "- SameState : 기업이 위치한 State과 은행이 위치한 State이 동일한지 여부(0=동일하지 않음, 1=동일함)\n",
    "- Recession : SBA 약정 발행 날짜가 글로벌 금융위기 시기에 포함되는지 여부(0=포함되지 않음, 1=포함)\n",
    "- ChgOffDate2 : 채무 불이행으로 선언되는 날짜(결측치 대체)\n",
    "- Term_re : 대출 기간 범주형 변수 (0=5년 미만, 1=5년 이상 15년 미만, 2=15년 이상)\n",
    "- smallmedium : 0=500명 미만(소기업), 1=500명 이상 1500명 이하(중기업)\n",
    "- createjob2 : 신규 직원 수 파생변수 0= 0 ~ 3인, 1= 4 ~ 20인, 2=21인 이상\n",
    "- retainedjob2 : 기존 직원 수 파생변수 0= 0 ~ 3인, 1= 4 ~ 20인, 2=21인 이상\n",
    "- dis_re : 0=1천만원 미만, 1=1천만원 이상 ~ 5천만원 미만, 2=5천만원 이상 ~ 1억 미만, 3=1억 이상 ~ 5억 미만, 4=5억 이상 ~ 10억 미만, 5=10억 이상\n",
    "- appv_pct : SBA_Appv / GrAppv\n",
    "- ~Year : 대금 지급일의 연도~\n",
    "- ~Month : 대금 지급일의 달~\n",
    "- ~DATE : 금리 데이터 날짜~\n",
    "- DPRIME : 금리 데이터 대출 금리\n",
    "- dprime2 : 0=저금리, 1=고금리(10% 이상)\n",
    "- ~Density : 산업 밀도~\n",
    "- ~Upper_Threshold : 성장 가능성 상~\n",
    "- ~Lower_Threshold : 성장 가능성 하~\n",
    "- ~Label : 성장 가능성 지표(1,2,3)~\n",
    "- growth_potential : 성장 가능성"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 라이브러리 불러오기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting tensorflow\n",
      "  Using cached tensorflow-2.13.1-cp38-cp38-win_amd64.whl (1.9 kB)\n",
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "ERROR: Could not find a version that satisfies the requirement tensorflow-intel==2.13.1; platform_system == \"Windows\" (from tensorflow) (from versions: 0.0.1, 2.10.0.dev20220728, 2.10.0rc0, 2.10.0rc1, 2.10.0rc2, 2.10.0rc3, 2.10.0, 2.10.1, 2.11.0rc0, 2.11.0rc1, 2.11.0rc2, 2.11.0, 2.11.1, 2.12.0rc0, 2.12.0rc1, 2.12.0, 2.12.1, 2.13.0rc0, 2.13.0rc1, 2.13.0rc2, 2.13.0)\n",
      "ERROR: No matching distribution found for tensorflow-intel==2.13.1; platform_system == \"Windows\" (from tensorflow)\n"
     ]
    }
   ],
   "source": [
    "#pip install --user tensorflow"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import geopandas as gpd\n",
    "\n",
    "%config InlineBackend.figure_format = 'retina'  #선명하게 만들기\n",
    "\n",
    "from sklearn.model_selection import train_test_split, GridSearchCV\n",
    "from sklearn.preprocessing import OneHotEncoder, StandardScaler, MinMaxScaler, MaxAbsScaler, RobustScaler\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score, precision_score, recall_score, f1_score, classification_report, roc_curve, auc\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn import svm\n",
    "from xgboost import XGBClassifier\n",
    "\n",
    "#import tensorflow as tf\n",
    "#from tensorflow.keras.models import Sequential\n",
    "#from tensorflow.keras.layers import Dense"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Warning 메세지를 뜨지 않게 해줌\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Window의 한글 폰트 설정\n",
    "plt.rc('font', family='Malgun Gothic')\n",
    "plt.rc('axes', unicode_minus=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 데이터 불러오기\n",
    "EDA 및 전처리가 완료된 df_final3.csv 파일 불러옴"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('df_final3.csv')\n",
    "df = data.copy() #원본 데이터"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(864747, 44)\n"
     ]
    },
    {
     "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>LoanNr_ChkDgt</th>\n",
       "      <th>Name</th>\n",
       "      <th>City</th>\n",
       "      <th>State</th>\n",
       "      <th>Zip</th>\n",
       "      <th>Bank</th>\n",
       "      <th>BankState</th>\n",
       "      <th>NAICS</th>\n",
       "      <th>ApprovalDate</th>\n",
       "      <th>ApprovalFY</th>\n",
       "      <th>...</th>\n",
       "      <th>Year</th>\n",
       "      <th>Month</th>\n",
       "      <th>DATE</th>\n",
       "      <th>DPRIME</th>\n",
       "      <th>dprime2</th>\n",
       "      <th>Density</th>\n",
       "      <th>Upper_Threshold</th>\n",
       "      <th>Lower_Threshold</th>\n",
       "      <th>Label</th>\n",
       "      <th>growth_potential</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000014003</td>\n",
       "      <td>ABC HOBBYCRAFT</td>\n",
       "      <td>EVANSVILLE</td>\n",
       "      <td>IN</td>\n",
       "      <td>47711</td>\n",
       "      <td>FIFTH THIRD BANK</td>\n",
       "      <td>OH</td>\n",
       "      <td>17</td>\n",
       "      <td>1997-02-28</td>\n",
       "      <td>1997</td>\n",
       "      <td>...</td>\n",
       "      <td>1999</td>\n",
       "      <td>2</td>\n",
       "      <td>1999-02-01</td>\n",
       "      <td>7.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0.172105</td>\n",
       "      <td>0.118196</td>\n",
       "      <td>0.055178</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1000935004</td>\n",
       "      <td>Hairline Inc.</td>\n",
       "      <td>Vancouver</td>\n",
       "      <td>WA</td>\n",
       "      <td>98663</td>\n",
       "      <td>WELLS FARGO BANK NATL ASSOC</td>\n",
       "      <td>SD</td>\n",
       "      <td>14</td>\n",
       "      <td>1997-04-08</td>\n",
       "      <td>1997</td>\n",
       "      <td>...</td>\n",
       "      <td>1999</td>\n",
       "      <td>2</td>\n",
       "      <td>1999-02-01</td>\n",
       "      <td>7.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0.090095</td>\n",
       "      <td>0.134566</td>\n",
       "      <td>0.055910</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1003535008</td>\n",
       "      <td>Jason A. Philbrook</td>\n",
       "      <td>Rockland</td>\n",
       "      <td>ME</td>\n",
       "      <td>4841</td>\n",
       "      <td>KEYBANK NATIONAL ASSOCIATION</td>\n",
       "      <td>OH</td>\n",
       "      <td>18</td>\n",
       "      <td>1999-01-11</td>\n",
       "      <td>1999</td>\n",
       "      <td>...</td>\n",
       "      <td>1999</td>\n",
       "      <td>2</td>\n",
       "      <td>1999-02-01</td>\n",
       "      <td>7.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0.041237</td>\n",
       "      <td>0.213874</td>\n",
       "      <td>0.051896</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1009994000</td>\n",
       "      <td>ARROW ENTERPRISE, INC.</td>\n",
       "      <td>BILLINGS</td>\n",
       "      <td>MT</td>\n",
       "      <td>59101</td>\n",
       "      <td>WESTERN SECURITY BANK</td>\n",
       "      <td>MT</td>\n",
       "      <td>5</td>\n",
       "      <td>1997-03-04</td>\n",
       "      <td>1997</td>\n",
       "      <td>...</td>\n",
       "      <td>1999</td>\n",
       "      <td>2</td>\n",
       "      <td>1999-02-01</td>\n",
       "      <td>7.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0.148380</td>\n",
       "      <td>0.209033</td>\n",
       "      <td>0.057460</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1011904005</td>\n",
       "      <td>ROBIN E. JONES, D.D.S.</td>\n",
       "      <td>TACOMA</td>\n",
       "      <td>WA</td>\n",
       "      <td>98404</td>\n",
       "      <td>KEYBANK NATIONAL ASSOCIATION</td>\n",
       "      <td>OR</td>\n",
       "      <td>8</td>\n",
       "      <td>1997-03-05</td>\n",
       "      <td>1997</td>\n",
       "      <td>...</td>\n",
       "      <td>1999</td>\n",
       "      <td>2</td>\n",
       "      <td>1999-02-01</td>\n",
       "      <td>7.75</td>\n",
       "      <td>0</td>\n",
       "      <td>0.094218</td>\n",
       "      <td>0.134566</td>\n",
       "      <td>0.055910</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 44 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   LoanNr_ChkDgt                    Name        City State    Zip  \\\n",
       "0     1000014003          ABC HOBBYCRAFT  EVANSVILLE    IN  47711   \n",
       "1     1000935004           Hairline Inc.   Vancouver    WA  98663   \n",
       "2     1003535008      Jason A. Philbrook    Rockland    ME   4841   \n",
       "3     1009994000  ARROW ENTERPRISE, INC.    BILLINGS    MT  59101   \n",
       "4     1011904005  ROBIN E. JONES, D.D.S.      TACOMA    WA  98404   \n",
       "\n",
       "                           Bank BankState  NAICS ApprovalDate  ApprovalFY  \\\n",
       "0              FIFTH THIRD BANK        OH     17   1997-02-28        1997   \n",
       "1   WELLS FARGO BANK NATL ASSOC        SD     14   1997-04-08        1997   \n",
       "2  KEYBANK NATIONAL ASSOCIATION        OH     18   1999-01-11        1999   \n",
       "3         WESTERN SECURITY BANK        MT      5   1997-03-04        1997   \n",
       "4  KEYBANK NATIONAL ASSOCIATION        OR      8   1997-03-05        1997   \n",
       "\n",
       "   ...  Year  Month        DATE  DPRIME  dprime2   Density  Upper_Threshold  \\\n",
       "0  ...  1999      2  1999-02-01    7.75        0  0.172105         0.118196   \n",
       "1  ...  1999      2  1999-02-01    7.75        0  0.090095         0.134566   \n",
       "2  ...  1999      2  1999-02-01    7.75        0  0.041237         0.213874   \n",
       "3  ...  1999      2  1999-02-01    7.75        0  0.148380         0.209033   \n",
       "4  ...  1999      2  1999-02-01    7.75        0  0.094218         0.134566   \n",
       "\n",
       "   Lower_Threshold Label growth_potential  \n",
       "0         0.055178     1                1  \n",
       "1         0.055910     3                2  \n",
       "2         0.051896     5                2  \n",
       "3         0.057460     3                1  \n",
       "4         0.055910     3                2  \n",
       "\n",
       "[5 rows x 44 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>LoanNr_ChkDgt</th>\n",
       "      <th>Name</th>\n",
       "      <th>City</th>\n",
       "      <th>State</th>\n",
       "      <th>Zip</th>\n",
       "      <th>Bank</th>\n",
       "      <th>BankState</th>\n",
       "      <th>NAICS</th>\n",
       "      <th>ApprovalDate</th>\n",
       "      <th>ApprovalFY</th>\n",
       "      <th>...</th>\n",
       "      <th>Year</th>\n",
       "      <th>Month</th>\n",
       "      <th>DATE</th>\n",
       "      <th>DPRIME</th>\n",
       "      <th>dprime2</th>\n",
       "      <th>Density</th>\n",
       "      <th>Upper_Threshold</th>\n",
       "      <th>Lower_Threshold</th>\n",
       "      <th>Label</th>\n",
       "      <th>growth_potential</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>864742</th>\n",
       "      <td>9022713002</td>\n",
       "      <td>BJORK, LAURA BEAUTY CONSULTANT</td>\n",
       "      <td>DALE CITY</td>\n",
       "      <td>VA</td>\n",
       "      <td>22193</td>\n",
       "      <td>MANUFACTURERS &amp; TRADERS TR CO</td>\n",
       "      <td>MD</td>\n",
       "      <td>1</td>\n",
       "      <td>1996-02-21</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>1996</td>\n",
       "      <td>2</td>\n",
       "      <td>1996-02-01</td>\n",
       "      <td>8.25</td>\n",
       "      <td>0</td>\n",
       "      <td>0.104331</td>\n",
       "      <td>0.139452</td>\n",
       "      <td>0.056099</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>864743</th>\n",
       "      <td>9035193000</td>\n",
       "      <td>PROFESSIONALS FOR NONPROFITS,I</td>\n",
       "      <td>NEW YORK</td>\n",
       "      <td>NY</td>\n",
       "      <td>10022</td>\n",
       "      <td>BANCO POPULAR NORTH AMERICA</td>\n",
       "      <td>NY</td>\n",
       "      <td>10</td>\n",
       "      <td>1996-02-26</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>1996</td>\n",
       "      <td>2</td>\n",
       "      <td>1996-02-01</td>\n",
       "      <td>8.25</td>\n",
       "      <td>0</td>\n",
       "      <td>0.094630</td>\n",
       "      <td>0.103363</td>\n",
       "      <td>0.066158</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>864744</th>\n",
       "      <td>9256453010</td>\n",
       "      <td>NORTHAIRE MARKET INC</td>\n",
       "      <td>SHAKOPEE</td>\n",
       "      <td>MN</td>\n",
       "      <td>55379</td>\n",
       "      <td>FIDELITY BANK</td>\n",
       "      <td>MN</td>\n",
       "      <td>20</td>\n",
       "      <td>1996-05-07</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>1996</td>\n",
       "      <td>2</td>\n",
       "      <td>1996-02-01</td>\n",
       "      <td>8.25</td>\n",
       "      <td>0</td>\n",
       "      <td>0.066200</td>\n",
       "      <td>0.133574</td>\n",
       "      <td>0.063179</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>864745</th>\n",
       "      <td>9459933001</td>\n",
       "      <td>COMPUTER RENAISSANCE</td>\n",
       "      <td>CARY</td>\n",
       "      <td>NC</td>\n",
       "      <td>27511</td>\n",
       "      <td>FIRST-CITIZENS BK &amp; TR CO</td>\n",
       "      <td>NC</td>\n",
       "      <td>17</td>\n",
       "      <td>1996-08-15</td>\n",
       "      <td>1996</td>\n",
       "      <td>...</td>\n",
       "      <td>1996</td>\n",
       "      <td>2</td>\n",
       "      <td>1996-02-01</td>\n",
       "      <td>8.25</td>\n",
       "      <td>0</td>\n",
       "      <td>0.187560</td>\n",
       "      <td>0.111641</td>\n",
       "      <td>0.060808</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>864746</th>\n",
       "      <td>7377094002</td>\n",
       "      <td>NORTHLAND DATACOMM CO</td>\n",
       "      <td>MINNEAPOLIS</td>\n",
       "      <td>MN</td>\n",
       "      <td>55311</td>\n",
       "      <td>BMO HARRIS BK NATL ASSOC</td>\n",
       "      <td>IL</td>\n",
       "      <td>20</td>\n",
       "      <td>2004-04-22</td>\n",
       "      <td>2004</td>\n",
       "      <td>...</td>\n",
       "      <td>2014</td>\n",
       "      <td>12</td>\n",
       "      <td>2014-12-01</td>\n",
       "      <td>3.25</td>\n",
       "      <td>0</td>\n",
       "      <td>0.066200</td>\n",
       "      <td>0.133574</td>\n",
       "      <td>0.063179</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 44 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        LoanNr_ChkDgt                            Name         City State  \\\n",
       "864742     9022713002  BJORK, LAURA BEAUTY CONSULTANT    DALE CITY    VA   \n",
       "864743     9035193000  PROFESSIONALS FOR NONPROFITS,I     NEW YORK    NY   \n",
       "864744     9256453010            NORTHAIRE MARKET INC     SHAKOPEE    MN   \n",
       "864745     9459933001            COMPUTER RENAISSANCE         CARY    NC   \n",
       "864746     7377094002           NORTHLAND DATACOMM CO  MINNEAPOLIS    MN   \n",
       "\n",
       "          Zip                           Bank BankState  NAICS ApprovalDate  \\\n",
       "864742  22193  MANUFACTURERS & TRADERS TR CO        MD      1   1996-02-21   \n",
       "864743  10022    BANCO POPULAR NORTH AMERICA        NY     10   1996-02-26   \n",
       "864744  55379                  FIDELITY BANK        MN     20   1996-05-07   \n",
       "864745  27511      FIRST-CITIZENS BK & TR CO        NC     17   1996-08-15   \n",
       "864746  55311       BMO HARRIS BK NATL ASSOC        IL     20   2004-04-22   \n",
       "\n",
       "        ApprovalFY  ...  Year  Month        DATE  DPRIME  dprime2   Density  \\\n",
       "864742        1996  ...  1996      2  1996-02-01    8.25        0  0.104331   \n",
       "864743        1996  ...  1996      2  1996-02-01    8.25        0  0.094630   \n",
       "864744        1996  ...  1996      2  1996-02-01    8.25        0  0.066200   \n",
       "864745        1996  ...  1996      2  1996-02-01    8.25        0  0.187560   \n",
       "864746        2004  ...  2014     12  2014-12-01    3.25        0  0.066200   \n",
       "\n",
       "        Upper_Threshold  Lower_Threshold Label growth_potential  \n",
       "864742         0.139452         0.056099     3                1  \n",
       "864743         0.103363         0.066158     3                1  \n",
       "864744         0.133574         0.063179     3                2  \n",
       "864745         0.111641         0.060808     1                1  \n",
       "864746         0.133574         0.063179     3                2  \n",
       "\n",
       "[5 rows x 44 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 864747 entries, 0 to 864746\n",
      "Data columns (total 44 columns):\n",
      " #   Column             Non-Null Count   Dtype  \n",
      "---  ------             --------------   -----  \n",
      " 0   LoanNr_ChkDgt      864747 non-null  int64  \n",
      " 1   Name               864747 non-null  object \n",
      " 2   City               864747 non-null  object \n",
      " 3   State              864747 non-null  object \n",
      " 4   Zip                864747 non-null  int64  \n",
      " 5   Bank               864747 non-null  object \n",
      " 6   BankState          864747 non-null  object \n",
      " 7   NAICS              864747 non-null  int64  \n",
      " 8   ApprovalDate       864747 non-null  object \n",
      " 9   ApprovalFY         864747 non-null  int64  \n",
      " 10  Term               864747 non-null  int64  \n",
      " 11  NoEmp              864747 non-null  int64  \n",
      " 12  NewExist           864747 non-null  int64  \n",
      " 13  CreateJob          864747 non-null  int64  \n",
      " 14  RetainedJob        864747 non-null  int64  \n",
      " 15  FranchiseCode      864747 non-null  int64  \n",
      " 16  UrbanRural         864747 non-null  int64  \n",
      " 17  LowDoc             864747 non-null  int64  \n",
      " 18  ChgOffDate         148743 non-null  object \n",
      " 19  DisbursementDate   864747 non-null  object \n",
      " 20  DisbursementGross  864747 non-null  int64  \n",
      " 21  MIS_Status         864747 non-null  int64  \n",
      " 22  ChgOffPrinGr       864747 non-null  int64  \n",
      " 23  GrAppv             864747 non-null  int64  \n",
      " 24  SBA_Appv           864747 non-null  int64  \n",
      " 25  SameState          864747 non-null  int64  \n",
      " 26  Recession          864747 non-null  int64  \n",
      " 27  ChgOffDate2        864747 non-null  object \n",
      " 28  Term_re            864747 non-null  int64  \n",
      " 29  smallmedium        864747 non-null  int64  \n",
      " 30  createjob2         864747 non-null  int64  \n",
      " 31  retainedjob2       864747 non-null  int64  \n",
      " 32  dis_re             864747 non-null  int64  \n",
      " 33  appv_pct           864747 non-null  float64\n",
      " 34  Year               864747 non-null  int64  \n",
      " 35  Month              864747 non-null  int64  \n",
      " 36  DATE               864747 non-null  object \n",
      " 37  DPRIME             864747 non-null  float64\n",
      " 38  dprime2            864747 non-null  int64  \n",
      " 39  Density            864747 non-null  float64\n",
      " 40  Upper_Threshold    864747 non-null  float64\n",
      " 41  Lower_Threshold    864747 non-null  float64\n",
      " 42  Label              864747 non-null  int64  \n",
      " 43  growth_potential   864747 non-null  int64  \n",
      "dtypes: float64(5), int64(29), object(10)\n",
      "memory usage: 290.3+ MB\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "print(df.shape)\n",
    "display(df.head())\n",
    "display(df.tail())\n",
    "print(df.info())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2차 전처리\n",
    "- 변수 선택\n",
    "- *범주형 데이터 수치화*\n",
    "- train, valid, test 셋 분리\n",
    "- 스케일링"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. 변수 선택"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 864747 entries, 0 to 864746\n",
      "Data columns (total 20 columns):\n",
      " #   Column            Non-Null Count   Dtype  \n",
      "---  ------            --------------   -----  \n",
      " 0   NAICS             864747 non-null  int64  \n",
      " 1   NoEmp             864747 non-null  int64  \n",
      " 2   NewExist          864747 non-null  int64  \n",
      " 3   FranchiseCode     864747 non-null  int64  \n",
      " 4   UrbanRural        864747 non-null  int64  \n",
      " 5   LowDoc            864747 non-null  int64  \n",
      " 6   MIS_Status        864747 non-null  int64  \n",
      " 7   GrAppv            864747 non-null  int64  \n",
      " 8   SBA_Appv          864747 non-null  int64  \n",
      " 9   SameState         864747 non-null  int64  \n",
      " 10  Recession         864747 non-null  int64  \n",
      " 11  Term_re           864747 non-null  int64  \n",
      " 12  smallmedium       864747 non-null  int64  \n",
      " 13  createjob2        864747 non-null  int64  \n",
      " 14  retainedjob2      864747 non-null  int64  \n",
      " 15  dis_re            864747 non-null  int64  \n",
      " 16  appv_pct          864747 non-null  float64\n",
      " 17  DPRIME            864747 non-null  float64\n",
      " 18  dprime2           864747 non-null  int64  \n",
      " 19  growth_potential  864747 non-null  int64  \n",
      "dtypes: float64(2), int64(18)\n",
      "memory usage: 132.0 MB\n"
     ]
    }
   ],
   "source": [
    "#필요없는 열은 제거\n",
    "df = df.drop([\"LoanNr_ChkDgt\",\"Name\",\"City\",\"State\",\"Zip\",\"Bank\",\"BankState\",\"ApprovalDate\",\"ApprovalFY\",\"Term\",\"CreateJob\",\"RetainedJob\",\"ChgOffDate\",\"DisbursementDate\",\"DisbursementGross\",\"ChgOffPrinGr\",\"ChgOffDate2\",\"Year\",\"Month\",\"DATE\",\"Density\",\"Upper_Threshold\",\"Lower_Threshold\",\"Label\"], axis=1)\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. train, valid, test 셋 분리"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1 test 셋 만들기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc  MIS_Status  \\\n",
      "825107     13      4         0              0           0       0           0   \n",
      "212849      8     16         0              0           0       0           0   \n",
      "724805      3      1         1              0           1       0           0   \n",
      "525131      1      7         1              0           1       0           0   \n",
      "256633     10     52         0              0           0       0           0   \n",
      "\n",
      "         GrAppv  SBA_Appv  SameState  Recession  Term_re  smallmedium  \\\n",
      "825107   306000    229500          1          0        2            0   \n",
      "212849   517000    387750          1          0        2            0   \n",
      "724805   338400    287640          0          0        1            0   \n",
      "525131   175300    131475          1          0        1            0   \n",
      "256633  1185000    749986          0          0        1            0   \n",
      "\n",
      "        createjob2  retainedjob2  dis_re   appv_pct  DPRIME  dprime2  \\\n",
      "825107           0             0       3  75.000000    8.50        0   \n",
      "212849           0             0       4  75.000000    4.75        0   \n",
      "724805           0             0       3  85.000000    6.00        0   \n",
      "525131           1             0       3  75.000000    8.25        0   \n",
      "256633           0             2       5  63.289958    8.50        0   \n",
      "\n",
      "        growth_potential  \n",
      "825107                 2  \n",
      "212849                 2  \n",
      "724805                 2  \n",
      "525131                 1  \n",
      "256633                 1  \n"
     ]
    }
   ],
   "source": [
    "# 시드 고정\n",
    "seed = 42\n",
    "df_sample = df.sample(n=1000, random_state=seed)\n",
    "\n",
    "# 샘플 데이터 확인\n",
    "print(df_sample.head())\n",
    "\n",
    "# 필요하다면 test 데이터 프레임으로 할당\n",
    "test = df_sample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1000, 20)\n"
     ]
    },
    {
     "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>NAICS</th>\n",
       "      <th>NoEmp</th>\n",
       "      <th>NewExist</th>\n",
       "      <th>FranchiseCode</th>\n",
       "      <th>UrbanRural</th>\n",
       "      <th>LowDoc</th>\n",
       "      <th>MIS_Status</th>\n",
       "      <th>GrAppv</th>\n",
       "      <th>SBA_Appv</th>\n",
       "      <th>SameState</th>\n",
       "      <th>Recession</th>\n",
       "      <th>Term_re</th>\n",
       "      <th>smallmedium</th>\n",
       "      <th>createjob2</th>\n",
       "      <th>retainedjob2</th>\n",
       "      <th>dis_re</th>\n",
       "      <th>appv_pct</th>\n",
       "      <th>DPRIME</th>\n",
       "      <th>dprime2</th>\n",
       "      <th>growth_potential</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>825107</th>\n",
       "      <td>13</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>306000</td>\n",
       "      <td>229500</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>75.000000</td>\n",
       "      <td>8.50</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212849</th>\n",
       "      <td>8</td>\n",
       "      <td>16</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>517000</td>\n",
       "      <td>387750</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>75.000000</td>\n",
       "      <td>4.75</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>724805</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>338400</td>\n",
       "      <td>287640</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>85.000000</td>\n",
       "      <td>6.00</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>525131</th>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>175300</td>\n",
       "      <td>131475</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>75.000000</td>\n",
       "      <td>8.25</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>256633</th>\n",
       "      <td>10</td>\n",
       "      <td>52</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1185000</td>\n",
       "      <td>749986</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>63.289958</td>\n",
       "      <td>8.50</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc  MIS_Status  \\\n",
       "825107     13      4         0              0           0       0           0   \n",
       "212849      8     16         0              0           0       0           0   \n",
       "724805      3      1         1              0           1       0           0   \n",
       "525131      1      7         1              0           1       0           0   \n",
       "256633     10     52         0              0           0       0           0   \n",
       "\n",
       "         GrAppv  SBA_Appv  SameState  Recession  Term_re  smallmedium  \\\n",
       "825107   306000    229500          1          0        2            0   \n",
       "212849   517000    387750          1          0        2            0   \n",
       "724805   338400    287640          0          0        1            0   \n",
       "525131   175300    131475          1          0        1            0   \n",
       "256633  1185000    749986          0          0        1            0   \n",
       "\n",
       "        createjob2  retainedjob2  dis_re   appv_pct  DPRIME  dprime2  \\\n",
       "825107           0             0       3  75.000000    8.50        0   \n",
       "212849           0             0       4  75.000000    4.75        0   \n",
       "724805           0             0       3  85.000000    6.00        0   \n",
       "525131           1             0       3  75.000000    8.25        0   \n",
       "256633           0             2       5  63.289958    8.50        0   \n",
       "\n",
       "        growth_potential  \n",
       "825107                 2  \n",
       "212849                 2  \n",
       "724805                 2  \n",
       "525131                 1  \n",
       "256633                 1  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>NAICS</th>\n",
       "      <th>NoEmp</th>\n",
       "      <th>NewExist</th>\n",
       "      <th>FranchiseCode</th>\n",
       "      <th>UrbanRural</th>\n",
       "      <th>LowDoc</th>\n",
       "      <th>MIS_Status</th>\n",
       "      <th>GrAppv</th>\n",
       "      <th>SBA_Appv</th>\n",
       "      <th>SameState</th>\n",
       "      <th>Recession</th>\n",
       "      <th>Term_re</th>\n",
       "      <th>smallmedium</th>\n",
       "      <th>createjob2</th>\n",
       "      <th>retainedjob2</th>\n",
       "      <th>dis_re</th>\n",
       "      <th>appv_pct</th>\n",
       "      <th>DPRIME</th>\n",
       "      <th>dprime2</th>\n",
       "      <th>growth_potential</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>451582</th>\n",
       "      <td>1</td>\n",
       "      <td>40</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>70000</td>\n",
       "      <td>35000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>4.00</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>613651</th>\n",
       "      <td>13</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>100000</td>\n",
       "      <td>80000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>80.0</td>\n",
       "      <td>8.25</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>582361</th>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>45000</td>\n",
       "      <td>22500</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>3.61</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>277365</th>\n",
       "      <td>14</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20000</td>\n",
       "      <td>10000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>50.0</td>\n",
       "      <td>9.50</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>316959</th>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>79000</td>\n",
       "      <td>79000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>100.0</td>\n",
       "      <td>8.50</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc  MIS_Status  \\\n",
       "451582      1     40         1              0           1       0           0   \n",
       "613651     13      4         0              0           1       1           0   \n",
       "582361      2      3         0              0           1       0           0   \n",
       "277365     14      8         0              0           1       0           0   \n",
       "316959      4      1         1              0           0       0           0   \n",
       "\n",
       "        GrAppv  SBA_Appv  SameState  Recession  Term_re  smallmedium  \\\n",
       "451582   70000     35000          0          0        0            0   \n",
       "613651  100000     80000          1          0        1            0   \n",
       "582361   45000     22500          1          0        1            0   \n",
       "277365   20000     10000          1          0        0            0   \n",
       "316959   79000     79000          0          0        2            0   \n",
       "\n",
       "        createjob2  retainedjob2  dis_re  appv_pct  DPRIME  dprime2  \\\n",
       "451582           0             2       2      50.0    4.00        0   \n",
       "613651           0             0       3      80.0    8.25        0   \n",
       "582361           0             1       2      50.0    3.61        0   \n",
       "277365           0             1       1      50.0    9.50        0   \n",
       "316959           2             0       2     100.0    8.50        0   \n",
       "\n",
       "        growth_potential  \n",
       "451582                 1  \n",
       "613651                 2  \n",
       "582361                 2  \n",
       "277365                 2  \n",
       "316959                 2  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 1000 entries, 825107 to 316959\n",
      "Data columns (total 20 columns):\n",
      " #   Column            Non-Null Count  Dtype  \n",
      "---  ------            --------------  -----  \n",
      " 0   NAICS             1000 non-null   int64  \n",
      " 1   NoEmp             1000 non-null   int64  \n",
      " 2   NewExist          1000 non-null   int64  \n",
      " 3   FranchiseCode     1000 non-null   int64  \n",
      " 4   UrbanRural        1000 non-null   int64  \n",
      " 5   LowDoc            1000 non-null   int64  \n",
      " 6   MIS_Status        1000 non-null   int64  \n",
      " 7   GrAppv            1000 non-null   int64  \n",
      " 8   SBA_Appv          1000 non-null   int64  \n",
      " 9   SameState         1000 non-null   int64  \n",
      " 10  Recession         1000 non-null   int64  \n",
      " 11  Term_re           1000 non-null   int64  \n",
      " 12  smallmedium       1000 non-null   int64  \n",
      " 13  createjob2        1000 non-null   int64  \n",
      " 14  retainedjob2      1000 non-null   int64  \n",
      " 15  dis_re            1000 non-null   int64  \n",
      " 16  appv_pct          1000 non-null   float64\n",
      " 17  DPRIME            1000 non-null   float64\n",
      " 18  dprime2           1000 non-null   int64  \n",
      " 19  growth_potential  1000 non-null   int64  \n",
      "dtypes: float64(2), int64(18)\n",
      "memory usage: 164.1 KB\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# test 데이터 구조 확인\n",
    "print(test.shape)\n",
    "display(test.head())\n",
    "display(test.tail())\n",
    "print(test.info())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc   GrAppv  \\\n",
      "825107     13      4         0              0           0       0   306000   \n",
      "212849      8     16         0              0           0       0   517000   \n",
      "724805      3      1         1              0           1       0   338400   \n",
      "525131      1      7         1              0           1       0   175300   \n",
      "256633     10     52         0              0           0       0  1185000   \n",
      "\n",
      "        SBA_Appv  SameState  Recession  Term_re  smallmedium  createjob2  \\\n",
      "825107    229500          1          0        2            0           0   \n",
      "212849    387750          1          0        2            0           0   \n",
      "724805    287640          0          0        1            0           0   \n",
      "525131    131475          1          0        1            0           1   \n",
      "256633    749986          0          0        1            0           0   \n",
      "\n",
      "        retainedjob2  dis_re   appv_pct  DPRIME  dprime2  growth_potential  \n",
      "825107             0       3  75.000000    8.50        0                 2  \n",
      "212849             0       4  75.000000    4.75        0                 2  \n",
      "724805             0       3  85.000000    6.00        0                 2  \n",
      "525131             0       3  75.000000    8.25        0                 1  \n",
      "256633             2       5  63.289958    8.50        0                 1  \n",
      "825107    0\n",
      "212849    0\n",
      "724805    0\n",
      "525131    0\n",
      "256633    0\n",
      "Name: MIS_Status, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# y_test에 MIS_Status 열 할당\n",
    "y_test = test['MIS_Status']\n",
    "\n",
    "# x_test에 MIS_Status를 제외한 열들 할당\n",
    "x_test = test.drop('MIS_Status', axis=1)\n",
    "\n",
    "# x_test 확인\n",
    "print(x_test.head())\n",
    "\n",
    "# y_test 확인\n",
    "print(y_test.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2 샘플링\n",
    "- 채무이행(0) : 채무불이행(1) = 5:5 비율로 샘플링하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1    4500\n",
      "0    4500\n",
      "Name: MIS_Status, dtype: int64\n"
     ]
    },
    {
     "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>NAICS</th>\n",
       "      <th>NoEmp</th>\n",
       "      <th>NewExist</th>\n",
       "      <th>FranchiseCode</th>\n",
       "      <th>UrbanRural</th>\n",
       "      <th>LowDoc</th>\n",
       "      <th>MIS_Status</th>\n",
       "      <th>GrAppv</th>\n",
       "      <th>SBA_Appv</th>\n",
       "      <th>SameState</th>\n",
       "      <th>Recession</th>\n",
       "      <th>Term_re</th>\n",
       "      <th>smallmedium</th>\n",
       "      <th>createjob2</th>\n",
       "      <th>retainedjob2</th>\n",
       "      <th>dis_re</th>\n",
       "      <th>appv_pct</th>\n",
       "      <th>DPRIME</th>\n",
       "      <th>dprime2</th>\n",
       "      <th>growth_potential</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>222395</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>25000</td>\n",
       "      <td>12500</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>50.0</td>\n",
       "      <td>7.50</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>439185</th>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>60000</td>\n",
       "      <td>30000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>4.00</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>273881</th>\n",
       "      <td>1</td>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>340000</td>\n",
       "      <td>255000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>75.0</td>\n",
       "      <td>4.84</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>138058</th>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>125000</td>\n",
       "      <td>93750</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>75.0</td>\n",
       "      <td>9.50</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>470427</th>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>50000</td>\n",
       "      <td>25000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>5.59</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>371085</th>\n",
       "      <td>17</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>10000</td>\n",
       "      <td>5000</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>50.0</td>\n",
       "      <td>4.01</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>751654</th>\n",
       "      <td>5</td>\n",
       "      <td>87</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>280000</td>\n",
       "      <td>238000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>85.0</td>\n",
       "      <td>6.00</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>601567</th>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>25000</td>\n",
       "      <td>12500</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>4.75</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>537593</th>\n",
       "      <td>14</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>20000</td>\n",
       "      <td>10000</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>50.0</td>\n",
       "      <td>7.74</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>460304</th>\n",
       "      <td>17</td>\n",
       "      <td>15</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>45000</td>\n",
       "      <td>22500</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>50.0</td>\n",
       "      <td>7.15</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9000 rows × 20 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc  MIS_Status  \\\n",
       "222395      1      2         0              0           0       0           0   \n",
       "439185      1      5         0              0           0       0           0   \n",
       "273881      1     17         1              1           0       0           0   \n",
       "138058      8      8         0              0           1       1           0   \n",
       "470427      8      1         1              0           0       0           0   \n",
       "...       ...    ...       ...            ...         ...     ...         ...   \n",
       "371085     17      3         0              0           0       0           1   \n",
       "751654      5     87         0              0           1       0           1   \n",
       "601567      2      4         0              0           1       0           1   \n",
       "537593     14      1         1              0           1       0           1   \n",
       "460304     17     15         0              0           0       0           1   \n",
       "\n",
       "        GrAppv  SBA_Appv  SameState  Recession  Term_re  smallmedium  \\\n",
       "222395   25000     12500          0          0        1            0   \n",
       "439185   60000     30000          1          0        0            0   \n",
       "273881  340000    255000          0          0        1            0   \n",
       "138058  125000     93750          1          0        1            0   \n",
       "470427   50000     25000          1          0        0            0   \n",
       "...        ...       ...        ...        ...      ...          ...   \n",
       "371085   10000      5000          0          0        0            0   \n",
       "751654  280000    238000          1          0        0            0   \n",
       "601567   25000     12500          0          0        0            0   \n",
       "537593   20000     10000          1          0        0            0   \n",
       "460304   45000     22500          0          0        0            0   \n",
       "\n",
       "        createjob2  retainedjob2  dis_re  appv_pct  DPRIME  dprime2  \\\n",
       "222395           0             0       1      50.0    7.50        0   \n",
       "439185           0             1       2      50.0    4.00        0   \n",
       "273881           1             1       3      75.0    4.84        0   \n",
       "138058           0             0       3      75.0    9.50        0   \n",
       "470427           1             1       2      50.0    5.59        0   \n",
       "...            ...           ...     ...       ...     ...      ...   \n",
       "371085           0             1       1      50.0    4.01        0   \n",
       "751654           0             0       3      85.0    6.00        0   \n",
       "601567           0             1       2      50.0    4.75        0   \n",
       "537593           0             0       1      50.0    7.74        0   \n",
       "460304           0             1       2      50.0    7.15        0   \n",
       "\n",
       "        growth_potential  \n",
       "222395                 1  \n",
       "439185                 1  \n",
       "273881                 1  \n",
       "138058                 2  \n",
       "470427                 2  \n",
       "...                  ...  \n",
       "371085                 1  \n",
       "751654                 1  \n",
       "601567                 2  \n",
       "537593                 2  \n",
       "460304                 1  \n",
       "\n",
       "[9000 rows x 20 columns]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 시드 고정\n",
    "seed = 42\n",
    "# df에서 test 데이터를 제외한 데이터를 추출\n",
    "df_remaining = df.drop(test.index)\n",
    "\n",
    "# MIS_Status가 0인 데이터와 1인 데이터로 분리\n",
    "df_0 = df_remaining[df_remaining['MIS_Status'] == 0]\n",
    "df_1 = df_remaining[df_remaining['MIS_Status'] == 1]\n",
    "\n",
    "# 0과 1의 비율 계산\n",
    "ratio_0 = len(df_0) / len(df_remaining)\n",
    "ratio_1 = len(df_1) / len(df_remaining)\n",
    "\n",
    "# 다운 샘플링할 데이터 개수 계산\n",
    "total_samples = 9000\n",
    "samples_0 = int(total_samples * 0.5)  # MIS_Status가 0인 샘플 개수\n",
    "samples_1 = int(total_samples * 0.5)  # MIS_Status가 1인 샘플 개수\n",
    "\n",
    "# 비율 조정을 위한 다운 샘플링\n",
    "df_0_downsampled = df_0.sample(n=samples_0, random_state=seed)\n",
    "df_1_downsampled = df_1.sample(n=samples_1, random_state=seed)\n",
    "\n",
    "# 다운 샘플링된 데이터 합치기\n",
    "df_downsampled = pd.concat([df_0_downsampled, df_1_downsampled])\n",
    "\n",
    "# 결과 데이터프레임 확인\n",
    "print(df_downsampled['MIS_Status'].value_counts())\n",
    "\n",
    "# 최종 데이터프레임 할당\n",
    "df_fin = df_downsampled\n",
    "df_fin"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### train, valid 데이터셋 만들기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train 데이터 비율:\n",
      "0    0.502286\n",
      "1    0.497714\n",
      "Name: MIS_Status, dtype: float64\n",
      "\n",
      "Valid 데이터 비율:\n",
      "1    0.508\n",
      "0    0.492\n",
      "Name: MIS_Status, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# train 데이터와 valid 데이터를 분할\n",
    "df_train, df_valid = train_test_split(df_fin, train_size=7000, test_size=2000, random_state=42)\n",
    "\n",
    "# 결과 확인\n",
    "print(\"Train 데이터 비율:\")\n",
    "print(df_train['MIS_Status'].value_counts(normalize=True))\n",
    "print(\"\\nValid 데이터 비율:\")\n",
    "print(df_valid['MIS_Status'].value_counts(normalize=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc  GrAppv  \\\n",
      "479937     20      1         0              0           0       0  100000   \n",
      "699666     17      1         0              0           1       0   26000   \n",
      "67684      10      5         0              0           0       0   50000   \n",
      "411530     17      2         0              0           0       0    6000   \n",
      "202436      8      4         0              0           0       0  125500   \n",
      "\n",
      "        SBA_Appv  SameState  Recession  Term_re  smallmedium  createjob2  \\\n",
      "479937     50000          1          0        1            0           0   \n",
      "699666     23400          1          0        1            0           0   \n",
      "67684      25000          1          0        0            0           0   \n",
      "411530      5100          0          0        0            0           0   \n",
      "202436     62750          0          0        1            0           0   \n",
      "\n",
      "        retainedjob2  dis_re  appv_pct  DPRIME  dprime2  growth_potential  \n",
      "479937             0       3      50.0    8.25        0                 2  \n",
      "699666             0       1      90.0   10.50        1                 1  \n",
      "67684              1       2      50.0    7.92        0                 1  \n",
      "411530             0       0      85.0    4.00        0                 1  \n",
      "202436             1       3      50.0    8.25        0                 2  \n",
      "479937    1\n",
      "699666    0\n",
      "67684     1\n",
      "411530    1\n",
      "202436    0\n",
      "Name: MIS_Status, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# df_train 분리\n",
    "y_train = df_train['MIS_Status']\n",
    "x_train = df_train.drop('MIS_Status', axis=1)\n",
    "\n",
    "# x_train, y_train 확인\n",
    "print(x_train.head())\n",
    "print(y_train.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "        NAICS  NoEmp  NewExist  FranchiseCode  UrbanRural  LowDoc  GrAppv  \\\n",
      "313128     14      3         0              0           0       1   50000   \n",
      "398709      8      6         0              0           0       0   40000   \n",
      "48061       8      1         1              0           0       0  500000   \n",
      "717593     16     14         0              0           1       0  445000   \n",
      "93573      20      3         0              0           1       0  100000   \n",
      "\n",
      "        SBA_Appv  SameState  Recession  Term_re  smallmedium  createjob2  \\\n",
      "313128     42500          0          0        0            0           0   \n",
      "398709     20000          0          0        1            0           0   \n",
      "48061     375000          1          0        1            0           0   \n",
      "717593    378250          1          0        1            0           0   \n",
      "93573      50000          0          0        0            0           0   \n",
      "\n",
      "        retainedjob2  dis_re  appv_pct  DPRIME  dprime2  growth_potential  \n",
      "313128             0       2      85.0    4.00        0                 2  \n",
      "398709             1       1      50.0    4.00        0                 2  \n",
      "48061              0       4      75.0    8.50        0                 2  \n",
      "717593             0       3      85.0    8.50        0                 2  \n",
      "93573              0       3      50.0    8.25        0                 2  \n",
      "313128    1\n",
      "398709    0\n",
      "48061     0\n",
      "717593    0\n",
      "93573     1\n",
      "Name: MIS_Status, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# df_valid 분리\n",
    "y_valid = df_valid['MIS_Status']\n",
    "x_valid = df_valid.drop('MIS_Status', axis=1)\n",
    "\n",
    "# x_valid, y_valid 확인\n",
    "print(x_valid.head())\n",
    "print(y_valid.head())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. 변수 스케일링"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(1) Standard Scaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train 데이터 스케일링 후 통계 정보:\n",
      "                 0             1             2             3             4   \\\n",
      "count  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03   \n",
      "mean  -2.100701e-16 -3.264373e-16  2.244554e-16  3.076943e-16 -3.976502e-16   \n",
      "std    1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00   \n",
      "min   -1.629067e+00 -3.253877e-01 -6.326769e-01 -2.325040e-01 -9.298805e-01   \n",
      "25%   -9.702496e-01 -2.867889e-01 -6.326769e-01 -2.325040e-01 -9.298805e-01   \n",
      "50%    3.473849e-01 -2.095914e-01 -6.326769e-01 -2.325040e-01 -9.298805e-01   \n",
      "75%    1.006202e+00 -1.659748e-02  1.580586e+00 -2.325040e-01  1.075407e+00   \n",
      "max    1.500315e+00  3.823479e+01  1.580586e+00  4.301001e+00  1.075407e+00   \n",
      "\n",
      "                 5             6             7             8             9   \\\n",
      "count  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03   \n",
      "mean  -1.158280e-16  5.443265e-17  1.140833e-16 -2.028536e-17  7.743726e-16   \n",
      "std    1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00   \n",
      "min   -3.489912e-01 -6.073070e-01 -5.724364e-01 -8.861483e-01 -2.335262e-01   \n",
      "25%   -3.489912e-01 -5.239696e-01 -5.094735e-01 -8.861483e-01 -2.335262e-01   \n",
      "50%   -3.489912e-01 -3.913874e-01 -3.844805e-01 -8.861483e-01 -2.335262e-01   \n",
      "75%   -3.489912e-01 -4.247308e-03  1.521712e-02  1.128479e+00 -2.335262e-01   \n",
      "max    2.865402e+00  1.157132e+01  1.442905e+01  1.128479e+00  4.282174e+00   \n",
      "\n",
      "                 10            11            12            13            14  \\\n",
      "count  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03   \n",
      "mean  -1.318311e-16  1.074454e-15  2.489691e-15  1.270254e-16 -2.591895e-16   \n",
      "std    1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00   \n",
      "min   -9.983395e-01 -1.690550e-02 -3.651971e-01 -6.300556e-01 -1.997465e+00   \n",
      "25%   -9.983395e-01 -1.690550e-02 -3.651971e-01 -6.300556e-01 -1.089114e+00   \n",
      "50%    4.760015e-01 -1.690550e-02 -3.651971e-01 -6.300556e-01 -1.807620e-01   \n",
      "75%    4.760015e-01 -1.690550e-02 -3.651971e-01  1.151200e+00  7.275896e-01   \n",
      "max    1.950343e+00  5.915235e+01  4.747562e+00  2.932456e+00  2.544293e+00   \n",
      "\n",
      "                 15            16            17            18  \n",
      "count  7.000000e+03  7.000000e+03  7.000000e+03  7.000000e+03  \n",
      "mean   1.941304e-16 -1.950820e-18  1.862875e-16  3.343675e-16  \n",
      "std    1.000071e+00  1.000071e+00  1.000071e+00  1.000071e+00  \n",
      "min   -3.468076e+00 -1.818063e+00 -1.172951e-01 -9.949222e-01  \n",
      "25%   -1.035014e+00 -9.825891e-01 -1.172951e-01 -9.949222e-01  \n",
      "50%    4.132366e-01  2.890133e-01 -1.172951e-01  9.074147e-01  \n",
      "75%    9.925369e-01  8.092143e-01 -1.172951e-01  9.074147e-01  \n",
      "max    1.861487e+00  2.516945e+00  8.525504e+00  2.809752e+00  \n"
     ]
    }
   ],
   "source": [
    "# 스케일러 초기화\n",
    "scaler = StandardScaler()\n",
    "\n",
    "# Train 데이터 스케일링\n",
    "x_train = scaler.fit_transform(x_train)\n",
    "\n",
    "# Valid 데이터 스케일링\n",
    "x_valid = scaler.transform(x_valid)\n",
    "\n",
    "# Test 데이터 스케일링\n",
    "x_test = scaler.transform(x_test)\n",
    "\n",
    "# 결과 확인\n",
    "print(\"Train 데이터 스케일링 후 통계 정보:\")\n",
    "print(pd.DataFrame(x_train).describe())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(2) MinMax Scaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train 데이터 스케일링 후 통계 정보:\n",
      "                0            1            2            3            4   \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.520571     0.008438     0.285857     0.051286     0.463714   \n",
      "std       0.319575     0.025935     0.451854     0.220596     0.498717   \n",
      "min       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "25%       0.210526     0.001001     0.000000     0.000000     0.000000   \n",
      "50%       0.631579     0.003003     0.000000     0.000000     0.000000   \n",
      "75%       0.842105     0.008008     1.000000     0.000000     1.000000   \n",
      "max       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
      "\n",
      "                5            6            7            8            9   \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.108571     0.049867     0.038159     0.439857     0.051714   \n",
      "std       0.311123     0.082117     0.066665     0.496405     0.221465   \n",
      "min       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "25%       0.000000     0.006843     0.004197     0.000000     0.000000   \n",
      "50%       0.000000     0.017729     0.012529     0.000000     0.000000   \n",
      "75%       0.000000     0.049518     0.039173     1.000000     0.000000   \n",
      "max       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
      "\n",
      "                10           11           12           13           14  \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.338571     0.000286     0.071429     0.176857     0.439800   \n",
      "std       0.339159     0.016902     0.195603     0.280721     0.220195   \n",
      "min       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "25%       0.000000     0.000000     0.000000     0.000000     0.200000   \n",
      "50%       0.500000     0.000000     0.000000     0.000000     0.400000   \n",
      "75%       0.500000     0.000000     0.000000     0.500000     0.600000   \n",
      "max       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
      "\n",
      "                15           16           17           18  \n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  \n",
      "mean      0.650724     0.419391     0.013571     0.261500  \n",
      "std       0.187646     0.230697     0.115712     0.262853  \n",
      "min       0.000000     0.000000     0.000000     0.000000  \n",
      "25%       0.456522     0.192727     0.000000     0.000000  \n",
      "50%       0.728261     0.486061     0.000000     0.500000  \n",
      "75%       0.836957     0.606061     0.000000     0.500000  \n",
      "max       1.000000     1.000000     1.000000     1.000000  \n"
     ]
    }
   ],
   "source": [
    "#객체 생성\n",
    "mms = MinMaxScaler()\n",
    "\n",
    "# Train 데이터 스케일링\n",
    "x_train = mms.fit_transform(x_train)\n",
    "\n",
    "# Valid 데이터 스케일링\n",
    "x_valid = mms.transform(x_valid)\n",
    "\n",
    "# Test 데이터 스케일링\n",
    "x_test = mms.transform(x_test)\n",
    "\n",
    "# 결과 확인\n",
    "print(\"Train 데이터 스케일링 후 통계 정보:\")\n",
    "print(pd.DataFrame(x_train).describe())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(3) MaxAbs Scaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train 데이터 스케일링 후 통계 정보:\n",
      "                0            1            2            3            4   \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.520571     0.008438     0.285857     0.051286     0.463714   \n",
      "std       0.319575     0.025935     0.451854     0.220596     0.498717   \n",
      "min       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "25%       0.210526     0.001001     0.000000     0.000000     0.000000   \n",
      "50%       0.631579     0.003003     0.000000     0.000000     0.000000   \n",
      "75%       0.842105     0.008008     1.000000     0.000000     1.000000   \n",
      "max       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
      "\n",
      "                5            6            7            8            9   \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.108571     0.049867     0.038159     0.439857     0.051714   \n",
      "std       0.311123     0.082117     0.066665     0.496405     0.221465   \n",
      "min       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "25%       0.000000     0.006843     0.004197     0.000000     0.000000   \n",
      "50%       0.000000     0.017729     0.012529     0.000000     0.000000   \n",
      "75%       0.000000     0.049518     0.039173     1.000000     0.000000   \n",
      "max       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
      "\n",
      "                10           11           12           13           14  \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.338571     0.000286     0.071429     0.176857     0.439800   \n",
      "std       0.339159     0.016902     0.195603     0.280721     0.220195   \n",
      "min       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "25%       0.000000     0.000000     0.000000     0.000000     0.200000   \n",
      "50%       0.500000     0.000000     0.000000     0.000000     0.400000   \n",
      "75%       0.500000     0.000000     0.000000     0.500000     0.600000   \n",
      "max       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
      "\n",
      "                15           16           17           18  \n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  \n",
      "mean      0.650724     0.419391     0.013571     0.261500  \n",
      "std       0.187646     0.230697     0.115712     0.262853  \n",
      "min       0.000000     0.000000     0.000000     0.000000  \n",
      "25%       0.456522     0.192727     0.000000     0.000000  \n",
      "50%       0.728261     0.486061     0.000000     0.500000  \n",
      "75%       0.836957     0.606061     0.000000     0.500000  \n",
      "max       1.000000     1.000000     1.000000     1.000000  \n"
     ]
    }
   ],
   "source": [
    "#객체 생성\n",
    "mas = MaxAbsScaler()\n",
    "\n",
    "# Train 데이터 스케일링\n",
    "x_train = mas.fit_transform(x_train)\n",
    "\n",
    "# Valid 데이터 스케일링\n",
    "x_valid = mas.transform(x_valid)\n",
    "\n",
    "# Test 데이터 스케일링\n",
    "x_test = mas.transform(x_test)\n",
    "\n",
    "# 결과 확인\n",
    "print(\"Train 데이터 스케일링 후 통계 정보:\")\n",
    "print(pd.DataFrame(x_train).describe())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(4) Robust Scaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train 데이터 스케일링 후 통계 정보:\n",
      "                0            1            2            3            4   \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean     -0.175762     0.775714     0.285857     0.051286     0.463714   \n",
      "std       0.505993     3.701344     0.451854     0.220596     0.498717   \n",
      "min      -1.000000    -0.428571     0.000000     0.000000     0.000000   \n",
      "25%      -0.666667    -0.285714     0.000000     0.000000     0.000000   \n",
      "50%       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "75%       0.333333     0.714286     1.000000     0.000000     1.000000   \n",
      "max       0.583333   142.285714     1.000000     1.000000     1.000000   \n",
      "\n",
      "                5            6            7            8            9   \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean      0.108571     0.753070     0.732776     0.439857     0.051714   \n",
      "std       0.311123     1.924242     1.906021     0.496405     0.221465   \n",
      "min       0.000000    -0.415452    -0.358222     0.000000     0.000000   \n",
      "25%       0.000000    -0.255102    -0.238222     0.000000     0.000000   \n",
      "50%       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "75%       0.000000     0.744898     0.761778     1.000000     0.000000   \n",
      "max       1.000000    23.017493    28.232889     1.000000     1.000000   \n",
      "\n",
      "                10           11           12           13           14  \\\n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  7000.000000   \n",
      "mean     -0.322857     0.000286     0.071429     0.353714     0.099500   \n",
      "std       0.678318     0.016902     0.195603     0.561442     0.550487   \n",
      "min      -1.000000     0.000000     0.000000     0.000000    -1.000000   \n",
      "25%      -1.000000     0.000000     0.000000     0.000000    -0.500000   \n",
      "50%       0.000000     0.000000     0.000000     0.000000     0.000000   \n",
      "75%       0.000000     0.000000     0.000000     1.000000     0.500000   \n",
      "max       1.000000     1.000000     1.000000     2.000000     1.500000   \n",
      "\n",
      "                15           16           17           18  \n",
      "count  7000.000000  7000.000000  7000.000000  7000.000000  \n",
      "mean     -0.203811    -0.161297     0.013571    -0.477000  \n",
      "std       0.493241     0.558137     0.115712     0.525707  \n",
      "min      -1.914286    -1.175953     0.000000    -1.000000  \n",
      "25%      -0.714286    -0.709677     0.000000    -1.000000  \n",
      "50%       0.000000     0.000000     0.000000     0.000000  \n",
      "75%       0.285714     0.290323     0.000000     0.000000  \n",
      "max       0.714286     1.243402     1.000000     1.000000  \n"
     ]
    }
   ],
   "source": [
    "#객체 생성\n",
    "rs = RobustScaler()\n",
    "\n",
    "# Train 데이터 스케일링\n",
    "x_train = rs.fit_transform(x_train)\n",
    "\n",
    "# Valid 데이터 스케일링\n",
    "x_valid = rs.transform(x_valid)\n",
    "\n",
    "# Test 데이터 스케일링\n",
    "x_test = rs.transform(x_test)\n",
    "\n",
    "# 결과 확인\n",
    "print(\"Train 데이터 스케일링 후 통계 정보:\")\n",
    "print(pd.DataFrame(x_train).describe())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 모델링\n",
    "- 로지스틱 회귀\n",
    "- 랜덤포레스트\n",
    "- **서포트 벡터 머신(SVM)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. 로지스틱 회귀"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.1 기본 모델"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 선언\n",
    "lg_model = LogisticRegression(random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-3 {color: black;}#sk-container-id-3 pre{padding: 0;}#sk-container-id-3 div.sk-toggleable {background-color: white;}#sk-container-id-3 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-3 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-3 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-3 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-3 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-3 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-3 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-3 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-3 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-3 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-3 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-3 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-3 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-3 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-3 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-3 div.sk-item {position: relative;z-index: 1;}#sk-container-id-3 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-3 div.sk-item::before, #sk-container-id-3 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-3 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-3 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-3 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-3 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-3 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-3 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-3 div.sk-label-container {text-align: center;}#sk-container-id-3 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-3 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-3\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>LogisticRegression(random_state=42)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-3\" type=\"checkbox\" checked><label for=\"sk-estimator-id-3\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">LogisticRegression</label><div class=\"sk-toggleable__content\"><pre>LogisticRegression(random_state=42)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "LogisticRegression(random_state=42)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 모델 학습\n",
    "lg_model.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 평가\n",
    "def model_eval(y_test, y_pred=None):\n",
    "    confusion = confusion_matrix(y_test, y_pred, labels=[True, False])\n",
    "    accuracy = accuracy_score(y_test, y_pred)\n",
    "    precision = precision_score(y_test, y_pred, labels=[True, False])\n",
    "    recall = recall_score(y_test, y_pred)\n",
    "    F1 = f1_score(y_test, y_pred, labels=[True, False])\n",
    "    print(\"오차행렬:\\n\", confusion)\n",
    "    print(\"\\n정확도: {:.4f}\".format(accuracy))\n",
    "    print(\"정밀도: {:.4f}\".format(precision))\n",
    "    print(\"재현율: {:.4f}\".format(recall))\n",
    "    print(\"F1: {:.4f}\".format(F1))\n",
    "    print(classification_report(y_test, y_pred, labels=[True, False]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[603 413]\n",
      " [351 633]]\n",
      "\n",
      "정확도: 0.6180\n",
      "정밀도: 0.6321\n",
      "재현율: 0.5935\n",
      "F1: 0.6122\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.63      0.59      0.61      1016\n",
      "       False       0.61      0.64      0.62       984\n",
      "\n",
      "    accuracy                           0.62      2000\n",
      "   macro avg       0.62      0.62      0.62      2000\n",
      "weighted avg       0.62      0.62      0.62      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pred=lg_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1.2 하이퍼파라미터 튜닝"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grid Search는 너무 오래 걸리므로, Manual Search 진행"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "param_grid = {\n",
    "    'C': [0.001, 0.01, 0.1, 1, 10, 100],        # 규제 강도를 조절하는 역수. 클수록 약한 규제.\n",
    "    'penalty': ['l1', 'l2'],                    # 규제 유형. 'l1'은 L1 규제, 'l2'는 L2 규제.\n",
    "    'solver': ['liblinear', 'saga'],            # 최적화에 사용할 알고리즘.\n",
    "    'max_iter': [100, 200, 300]                 # 최대 반복 횟수.\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[603 413]\n",
      " [351 633]]\n",
      "\n",
      "정확도: 0.6180\n",
      "정밀도: 0.6321\n",
      "재현율: 0.5935\n",
      "F1: 0.6122\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.63      0.59      0.61      1016\n",
      "       False       0.61      0.64      0.62       984\n",
      "\n",
      "    accuracy                           0.62      2000\n",
      "   macro avg       0.62      0.62      0.62      2000\n",
      "weighted avg       0.62      0.62      0.62      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = LogisticRegression(random_state=42, C=10, penalty='l1', solver='saga', max_iter=300)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. 랜덤 포레스트(Random Forest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.1 기본 모델"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 선언\n",
    "rf_model = RandomForestClassifier(random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-1 {color: black;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>RandomForestClassifier(random_state=42)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" checked><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">RandomForestClassifier</label><div class=\"sk-toggleable__content\"><pre>RandomForestClassifier(random_state=42)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "RandomForestClassifier(random_state=42)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 모델 학습\n",
    "rf_model.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[734 282]\n",
      " [184 800]]\n",
      "\n",
      "정확도: 0.7670\n",
      "정밀도: 0.7996\n",
      "재현율: 0.7224\n",
      "F1: 0.7590\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.72      0.76      1016\n",
      "       False       0.74      0.81      0.77       984\n",
      "\n",
      "    accuracy                           0.77      2000\n",
      "   macro avg       0.77      0.77      0.77      2000\n",
      "weighted avg       0.77      0.77      0.77      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pred=rf_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.2 하이퍼파라미터 튜닝"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grid Search는 너무 오래 걸리므로, Manual Search 진행"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "param_grid = {\n",
    "    'n_estimators': [50, 100, 150],  # 트리의 개수\n",
    "    'max_depth': [None, 10, 20, 30],   # 트리의 최대 깊이\n",
    "    'min_samples_split': [2, 5, 10],   # 내부 노드를 분할하기 위한 최소 샘플 수\n",
    "    'min_samples_leaf': [1, 2, 4]      # 리프 노드에 필요한 최소 샘플 수\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[731 285]\n",
      " [165 819]]\n",
      "\n",
      "정확도: 0.7750\n",
      "정밀도: 0.8158\n",
      "재현율: 0.7195\n",
      "F1: 0.7646\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.82      0.72      0.76      1016\n",
      "       False       0.74      0.83      0.78       984\n",
      "\n",
      "    accuracy                           0.78      2000\n",
      "   macro avg       0.78      0.78      0.77      2000\n",
      "weighted avg       0.78      0.78      0.77      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = RandomForestClassifier(random_state=42, n_estimators=50, max_depth=None, min_samples_split=2, min_samples_leaf=2)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[734 282]\n",
      " [148 836]]\n",
      "\n",
      "정확도: 0.7850\n",
      "정밀도: 0.8322\n",
      "재현율: 0.7224\n",
      "F1: 0.7734\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.83      0.72      0.77      1016\n",
      "       False       0.75      0.85      0.80       984\n",
      "\n",
      "    accuracy                           0.79      2000\n",
      "   macro avg       0.79      0.79      0.78      2000\n",
      "weighted avg       0.79      0.79      0.78      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = RandomForestClassifier(random_state=42, n_estimators=50, max_depth=None, min_samples_split=2, min_samples_leaf=4)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[744 272]\n",
      " [161 823]]\n",
      "\n",
      "정확도: 0.7835\n",
      "정밀도: 0.8221\n",
      "재현율: 0.7323\n",
      "F1: 0.7746\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.82      0.73      0.77      1016\n",
      "       False       0.75      0.84      0.79       984\n",
      "\n",
      "    accuracy                           0.78      2000\n",
      "   macro avg       0.79      0.78      0.78      2000\n",
      "weighted avg       0.79      0.78      0.78      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = RandomForestClassifier(random_state=42, n_estimators=50, max_depth=20, min_samples_split=5, min_samples_leaf=4)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[732 284]\n",
      " [164 820]]\n",
      "\n",
      "정확도: 0.7760\n",
      "정밀도: 0.8170\n",
      "재현율: 0.7205\n",
      "F1: 0.7657\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.82      0.72      0.77      1016\n",
      "       False       0.74      0.83      0.79       984\n",
      "\n",
      "    accuracy                           0.78      2000\n",
      "   macro avg       0.78      0.78      0.78      2000\n",
      "weighted avg       0.78      0.78      0.78      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = RandomForestClassifier(random_state=42, n_estimators=150, max_depth=None, min_samples_split=2, min_samples_leaf=2)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[736 280]\n",
      " [148 836]]\n",
      "\n",
      "정확도: 0.7860\n",
      "정밀도: 0.8326\n",
      "재현율: 0.7244\n",
      "F1: 0.7747\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.83      0.72      0.77      1016\n",
      "       False       0.75      0.85      0.80       984\n",
      "\n",
      "    accuracy                           0.79      2000\n",
      "   macro avg       0.79      0.79      0.79      2000\n",
      "weighted avg       0.79      0.79      0.79      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = RandomForestClassifier(random_state=42, n_estimators=150, max_depth=30, min_samples_split=5, min_samples_leaf=4)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "=> 0.7747이 제일 높은 것으로 나타남."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2.3 변수 중요도"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 373,
       "width": 543
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 변수 중요도 계산\n",
    "\n",
    "importances_values = pd.Series(ms_model.feature_importances_, index=x_valid.columns).sort_values(ascending=False)\n",
    "\n",
    "# 변수 중요도 시각화\n",
    "\n",
    "plt.figure(figsize=(8,6))\n",
    "plt.title('Feature Importances')\n",
    "sns.barplot(x=importances_values, y=importances_values.index)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. 서포트 벡터 머신(SVM)\n",
    "- 전통적인 이진분류 머신러닝 기법\n",
    "- 과적합을 피할 수 있음\n",
    "- 데이터셋이 작아도 가능"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1 기본 모델"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 선언\n",
    "svm_model = svm.SVC(random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-1 {color: black;}#sk-container-id-1 pre{padding: 0;}#sk-container-id-1 div.sk-toggleable {background-color: white;}#sk-container-id-1 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-1 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-1 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-1 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-1 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-1 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-1 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-1 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-1 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-1 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-1 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-1 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-1 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-1 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-1 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-1 div.sk-item {position: relative;z-index: 1;}#sk-container-id-1 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-1 div.sk-item::before, #sk-container-id-1 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-1 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-1 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-1 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-1 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-1 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-1 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-1 div.sk-label-container {text-align: center;}#sk-container-id-1 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-1 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-1\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>SVC(random_state=42)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-1\" type=\"checkbox\" checked><label for=\"sk-estimator-id-1\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">SVC</label><div class=\"sk-toggleable__content\"><pre>SVC(random_state=42)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "SVC(random_state=42)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 모델 학습\n",
    "svm_model.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[695 321]\n",
      " [202 782]]\n",
      "\n",
      "정확도: 0.7385\n",
      "정밀도: 0.7748\n",
      "재현율: 0.6841\n",
      "F1: 0.7266\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.77      0.68      0.73      1016\n",
      "       False       0.71      0.79      0.75       984\n",
      "\n",
      "    accuracy                           0.74      2000\n",
      "   macro avg       0.74      0.74      0.74      2000\n",
      "weighted avg       0.74      0.74      0.74      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pred=svm_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "=> f1-score가 약 0.73"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2 하이퍼파라미터 튜닝\n",
    "- svm 모델에서 특성 중요도는 kernel=linear일 때만 가능\n",
    "- C : 오류를 얼마나 허용할 것인지 클수록 하드마진, 작을수록 소프트마진에 가까움\n",
    "- kernel : rbf(가우시안 커널), linear, poly, sigmoid, precomputed\n",
    "- degree : 다항식 커널의 차수 결정(default=3)\n",
    "- gamma : 결정경계를 얼마나 유연하게 그릴지 결정함. 클수록 오버피팅 발생 가능성 높아짐(default='scale')\n",
    "- coef0 : 다항식 커널에 있는 상수항 r(default=0.0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "kernel='linear'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[703 313]\n",
      " [237 747]]\n",
      "\n",
      "정확도: 0.7250\n",
      "정밀도: 0.7479\n",
      "재현율: 0.6919\n",
      "F1: 0.7188\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.75      0.69      0.72      1016\n",
      "       False       0.70      0.76      0.73       984\n",
      "\n",
      "    accuracy                           0.73      2000\n",
      "   macro avg       0.73      0.73      0.72      2000\n",
      "weighted avg       0.73      0.72      0.72      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# 모델 선언\n",
    "svm_model = svm.SVC(random_state=42, kernel='linear')\n",
    "\n",
    "# 모델 학습\n",
    "svm_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred=svm_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "GridSearch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Parameters: {'C': 10, 'gamma': 'auto', 'kernel': 'rbf'}\n"
     ]
    }
   ],
   "source": [
    "# SVM 모델 정의\n",
    "svm_model = svm.SVC(random_state=42)\n",
    "\n",
    "# 그리드 서치를 위한 하이퍼파라미터 후보군 설정\n",
    "param_grid = {\n",
    "    'C': [0.1, 1, 10],\n",
    "    'kernel': ['linear', 'rbf', 'poly'],\n",
    "    'gamma': ['scale', 'auto']\n",
    "}\n",
    "\n",
    "# 그리드 서치 객체 생성\n",
    "grid_search = GridSearchCV(estimator=svm_model, param_grid=param_grid, cv=5, scoring='accuracy')\n",
    "\n",
    "# 그리드 서치 수행\n",
    "grid_search.fit(x_train, y_train)\n",
    "\n",
    "# 최적 하이퍼파라미터 출력\n",
    "print(\"Best Parameters:\", grid_search.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[715 301]\n",
      " [180 804]]\n",
      "\n",
      "정확도: 0.7595\n",
      "정밀도: 0.7989\n",
      "재현율: 0.7037\n",
      "F1: 0.7483\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.70      0.75      1016\n",
      "       False       0.73      0.82      0.77       984\n",
      "\n",
      "    accuracy                           0.76      2000\n",
      "   macro avg       0.76      0.76      0.76      2000\n",
      "weighted avg       0.76      0.76      0.76      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# 최적 하이퍼파라미터를 사용한 모델 정의\n",
    "gs_model = svm.SVC(random_state=42, C=10, kernel='rbf', gamma='auto')\n",
    "\n",
    "# 모델 학습\n",
    "gs_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = gs_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "=> f1-score가 0.7483으로, 가장 높은 수치를 보여줌."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Manual Search를 통해 수동으로 하이퍼파라미터 조정하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[712 304]\n",
      " [180 804]]\n",
      "\n",
      "정확도: 0.7580\n",
      "정밀도: 0.7982\n",
      "재현율: 0.7008\n",
      "F1: 0.7463\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.70      0.75      1016\n",
      "       False       0.73      0.82      0.77       984\n",
      "\n",
      "    accuracy                           0.76      2000\n",
      "   macro avg       0.76      0.76      0.76      2000\n",
      "weighted avg       0.76      0.76      0.76      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = svm.SVC(random_state=42, C=9, kernel='rbf', gamma='auto')\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[710 306]\n",
      " [183 801]]\n",
      "\n",
      "정확도: 0.7555\n",
      "정밀도: 0.7951\n",
      "재현율: 0.6988\n",
      "F1: 0.7438\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.70      0.74      1016\n",
      "       False       0.72      0.81      0.77       984\n",
      "\n",
      "    accuracy                           0.76      2000\n",
      "   macro avg       0.76      0.76      0.75      2000\n",
      "weighted avg       0.76      0.76      0.75      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = svm.SVC(random_state=42, C=8, kernel='rbf', gamma='auto')\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[719 297]\n",
      " [189 795]]\n",
      "\n",
      "정확도: 0.7570\n",
      "정밀도: 0.7919\n",
      "재현율: 0.7077\n",
      "F1: 0.7474\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.79      0.71      0.75      1016\n",
      "       False       0.73      0.81      0.77       984\n",
      "\n",
      "    accuracy                           0.76      2000\n",
      "   macro avg       0.76      0.76      0.76      2000\n",
      "weighted avg       0.76      0.76      0.76      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = svm.SVC(random_state=42, C=10, kernel='rbf', gamma=0.1)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[732 284]\n",
      " [257 727]]\n",
      "\n",
      "정확도: 0.7295\n",
      "정밀도: 0.7401\n",
      "재현율: 0.7205\n",
      "F1: 0.7302\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.74      0.72      0.73      1016\n",
      "       False       0.72      0.74      0.73       984\n",
      "\n",
      "    accuracy                           0.73      2000\n",
      "   macro avg       0.73      0.73      0.73      2000\n",
      "weighted avg       0.73      0.73      0.73      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = svm.SVC(random_state=42, C=10, kernel='rbf', gamma=0.5)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grid Search로 찾은 모델의 f1-score가 가장 높으므로, gs_model로 추후 분석 진행!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.3 시각화"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "ROC 곡선"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 388,
       "width": 502
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 예측된 확률 계산\n",
    "probs = gs_model.decision_function(x_valid)\n",
    "\n",
    "# ROC 곡선 계산\n",
    "fpr, tpr, thresholds = roc_curve(y_valid, probs)\n",
    "roc_auc = auc(fpr, tpr)\n",
    "\n",
    "# ROC 곡선 그리기\n",
    "plt.figure(figsize=(8, 6))\n",
    "plt.plot(fpr, tpr, color='darkorange', lw=2, label='ROC curve (area = %0.2f)' % roc_auc)\n",
    "plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.05])\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.title('Receiver Operating Characteristic (ROC) Curve')\n",
    "plt.legend(loc=\"lower right\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "=> AUC = 0.83"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. XGBoost"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.1 기본 모델"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 선언\n",
    "xg_model = XGBClassifier(random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style>#sk-container-id-4 {color: black;}#sk-container-id-4 pre{padding: 0;}#sk-container-id-4 div.sk-toggleable {background-color: white;}#sk-container-id-4 label.sk-toggleable__label {cursor: pointer;display: block;width: 100%;margin-bottom: 0;padding: 0.3em;box-sizing: border-box;text-align: center;}#sk-container-id-4 label.sk-toggleable__label-arrow:before {content: \"▸\";float: left;margin-right: 0.25em;color: #696969;}#sk-container-id-4 label.sk-toggleable__label-arrow:hover:before {color: black;}#sk-container-id-4 div.sk-estimator:hover label.sk-toggleable__label-arrow:before {color: black;}#sk-container-id-4 div.sk-toggleable__content {max-height: 0;max-width: 0;overflow: hidden;text-align: left;background-color: #f0f8ff;}#sk-container-id-4 div.sk-toggleable__content pre {margin: 0.2em;color: black;border-radius: 0.25em;background-color: #f0f8ff;}#sk-container-id-4 input.sk-toggleable__control:checked~div.sk-toggleable__content {max-height: 200px;max-width: 100%;overflow: auto;}#sk-container-id-4 input.sk-toggleable__control:checked~label.sk-toggleable__label-arrow:before {content: \"▾\";}#sk-container-id-4 div.sk-estimator input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-4 div.sk-label input.sk-toggleable__control:checked~label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-4 input.sk-hidden--visually {border: 0;clip: rect(1px 1px 1px 1px);clip: rect(1px, 1px, 1px, 1px);height: 1px;margin: -1px;overflow: hidden;padding: 0;position: absolute;width: 1px;}#sk-container-id-4 div.sk-estimator {font-family: monospace;background-color: #f0f8ff;border: 1px dotted black;border-radius: 0.25em;box-sizing: border-box;margin-bottom: 0.5em;}#sk-container-id-4 div.sk-estimator:hover {background-color: #d4ebff;}#sk-container-id-4 div.sk-parallel-item::after {content: \"\";width: 100%;border-bottom: 1px solid gray;flex-grow: 1;}#sk-container-id-4 div.sk-label:hover label.sk-toggleable__label {background-color: #d4ebff;}#sk-container-id-4 div.sk-serial::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: 0;}#sk-container-id-4 div.sk-serial {display: flex;flex-direction: column;align-items: center;background-color: white;padding-right: 0.2em;padding-left: 0.2em;position: relative;}#sk-container-id-4 div.sk-item {position: relative;z-index: 1;}#sk-container-id-4 div.sk-parallel {display: flex;align-items: stretch;justify-content: center;background-color: white;position: relative;}#sk-container-id-4 div.sk-item::before, #sk-container-id-4 div.sk-parallel-item::before {content: \"\";position: absolute;border-left: 1px solid gray;box-sizing: border-box;top: 0;bottom: 0;left: 50%;z-index: -1;}#sk-container-id-4 div.sk-parallel-item {display: flex;flex-direction: column;z-index: 1;position: relative;background-color: white;}#sk-container-id-4 div.sk-parallel-item:first-child::after {align-self: flex-end;width: 50%;}#sk-container-id-4 div.sk-parallel-item:last-child::after {align-self: flex-start;width: 50%;}#sk-container-id-4 div.sk-parallel-item:only-child::after {width: 0;}#sk-container-id-4 div.sk-dashed-wrapped {border: 1px dashed gray;margin: 0 0.4em 0.5em 0.4em;box-sizing: border-box;padding-bottom: 0.4em;background-color: white;}#sk-container-id-4 div.sk-label label {font-family: monospace;font-weight: bold;display: inline-block;line-height: 1.2em;}#sk-container-id-4 div.sk-label-container {text-align: center;}#sk-container-id-4 div.sk-container {/* jupyter's `normalize.less` sets `[hidden] { display: none; }` but bootstrap.min.css set `[hidden] { display: none !important; }` so we also need the `!important` here to be able to override the default hidden behavior on the sphinx rendered scikit-learn.org. See: https://github.com/scikit-learn/scikit-learn/issues/21755 */display: inline-block !important;position: relative;}#sk-container-id-4 div.sk-text-repr-fallback {display: none;}</style><div id=\"sk-container-id-4\" class=\"sk-top-container\"><div class=\"sk-text-repr-fallback\"><pre>XGBClassifier(base_score=None, booster=None, callbacks=None,\n",
       "              colsample_bylevel=None, colsample_bynode=None,\n",
       "              colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "              enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "              gamma=None, grow_policy=None, importance_type=None,\n",
       "              interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "              max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "              max_delta_step=None, max_depth=None, max_leaves=None,\n",
       "              min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "              multi_strategy=None, n_estimators=None, n_jobs=None,\n",
       "              num_parallel_tree=None, random_state=42, ...)</pre><b>In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook. <br />On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.</b></div><div class=\"sk-container\" hidden><div class=\"sk-item\"><div class=\"sk-estimator sk-toggleable\"><input class=\"sk-toggleable__control sk-hidden--visually\" id=\"sk-estimator-id-4\" type=\"checkbox\" checked><label for=\"sk-estimator-id-4\" class=\"sk-toggleable__label sk-toggleable__label-arrow\">XGBClassifier</label><div class=\"sk-toggleable__content\"><pre>XGBClassifier(base_score=None, booster=None, callbacks=None,\n",
       "              colsample_bylevel=None, colsample_bynode=None,\n",
       "              colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "              enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "              gamma=None, grow_policy=None, importance_type=None,\n",
       "              interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "              max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "              max_delta_step=None, max_depth=None, max_leaves=None,\n",
       "              min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "              multi_strategy=None, n_estimators=None, n_jobs=None,\n",
       "              num_parallel_tree=None, random_state=42, ...)</pre></div></div></div></div></div>"
      ],
      "text/plain": [
       "XGBClassifier(base_score=None, booster=None, callbacks=None,\n",
       "              colsample_bylevel=None, colsample_bynode=None,\n",
       "              colsample_bytree=None, device=None, early_stopping_rounds=None,\n",
       "              enable_categorical=False, eval_metric=None, feature_types=None,\n",
       "              gamma=None, grow_policy=None, importance_type=None,\n",
       "              interaction_constraints=None, learning_rate=None, max_bin=None,\n",
       "              max_cat_threshold=None, max_cat_to_onehot=None,\n",
       "              max_delta_step=None, max_depth=None, max_leaves=None,\n",
       "              min_child_weight=None, missing=nan, monotone_constraints=None,\n",
       "              multi_strategy=None, n_estimators=None, n_jobs=None,\n",
       "              num_parallel_tree=None, random_state=42, ...)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 모델 학습\n",
    "xg_model.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[754 262]\n",
      " [208 776]]\n",
      "\n",
      "정확도: 0.7650\n",
      "정밀도: 0.7838\n",
      "재현율: 0.7421\n",
      "F1: 0.7624\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.78      0.74      0.76      1016\n",
      "       False       0.75      0.79      0.77       984\n",
      "\n",
      "    accuracy                           0.77      2000\n",
      "   macro avg       0.77      0.77      0.76      2000\n",
      "weighted avg       0.77      0.77      0.76      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# 모델 평가\n",
    "pred=xg_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2 하이퍼파라미터 튜닝"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Grid Search는 너무 오래 걸리므로, Manual Search 진행"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "    'learning_rate': [0.01, 0.1, 0.2],                # 학습률\n",
    "    'n_estimators': [100, 200, 300],                  # 트리의 개수\n",
    "    'max_depth': [3, 4, 5],                           # 트리의 최대 깊이\n",
    "    'subsample': [0.8, 0.9, 1.0],                     # 트리를 훈련하는 데 사용되는 샘플 비율\n",
    "    'colsample_bytree': [0.8, 0.9, 1.0],              # 각 트리마다 사용되는 특성(feature)의 비율\n",
    "    'gamma': [0, 0.1, 0.2],                           # 분할을 위한 최소 손실 감소\n",
    "    'reg_alpha': [0, 0.1, 0.2],                       # L1 규제\n",
    "    'reg_lambda': [0, 0.1, 0.2]                       # L2 규제\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[738 278]\n",
      " [177 807]]\n",
      "\n",
      "정확도: 0.7725\n",
      "정밀도: 0.8066\n",
      "재현율: 0.7264\n",
      "F1: 0.7644\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.81      0.73      0.76      1016\n",
      "       False       0.74      0.82      0.78       984\n",
      "\n",
      "    accuracy                           0.77      2000\n",
      "   macro avg       0.78      0.77      0.77      2000\n",
      "weighted avg       0.78      0.77      0.77      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = XGBClassifier(random_state=42, learning_rate=0.1, n_estimators=100, max_depth=3, subsample=0.8, colsample_bytree=0.8,\n",
    "                        gamma=0, reg_alpha=0, reg_lambda=0)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[759 257]\n",
      " [186 798]]\n",
      "\n",
      "정확도: 0.7785\n",
      "정밀도: 0.8032\n",
      "재현율: 0.7470\n",
      "F1: 0.7741\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.75      0.77      1016\n",
      "       False       0.76      0.81      0.78       984\n",
      "\n",
      "    accuracy                           0.78      2000\n",
      "   macro avg       0.78      0.78      0.78      2000\n",
      "weighted avg       0.78      0.78      0.78      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = XGBClassifier(random_state=42, learning_rate=0.2, n_estimators=100, max_depth=3, subsample=0.8, colsample_bytree=0.8,\n",
    "                        gamma=0, reg_alpha=0, reg_lambda=0)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[768 248]\n",
      " [195 789]]\n",
      "\n",
      "정확도: 0.7785\n",
      "정밀도: 0.7975\n",
      "재현율: 0.7559\n",
      "F1: 0.7761\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.76      0.78      1016\n",
      "       False       0.76      0.80      0.78       984\n",
      "\n",
      "    accuracy                           0.78      2000\n",
      "   macro avg       0.78      0.78      0.78      2000\n",
      "weighted avg       0.78      0.78      0.78      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = XGBClassifier(random_state=42, learning_rate=0.2, n_estimators=100, max_depth=3, subsample=0.8, colsample_bytree=0.9,\n",
    "                        gamma=0.1, reg_alpha=0, reg_lambda=0)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "오차행렬:\n",
      " [[768 248]\n",
      " [195 789]]\n",
      "\n",
      "정확도: 0.7785\n",
      "정밀도: 0.7975\n",
      "재현율: 0.7559\n",
      "F1: 0.7761\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "        True       0.80      0.76      0.78      1016\n",
      "       False       0.76      0.80      0.78       984\n",
      "\n",
      "    accuracy                           0.78      2000\n",
      "   macro avg       0.78      0.78      0.78      2000\n",
      "weighted avg       0.78      0.78      0.78      2000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#모델 정의\n",
    "ms_model = XGBClassifier(random_state=42, learning_rate=0.2, n_estimators=100, max_depth=3, subsample=0.8, colsample_bytree=0.9,\n",
    "                        gamma=0, reg_alpha=0, reg_lambda=0)\n",
    "\n",
    "# 모델 학습\n",
    "ms_model.fit(x_train, y_train)\n",
    "\n",
    "# 모델 평가\n",
    "pred = ms_model.predict(x_valid)\n",
    "model_eval(y_valid, pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.3 변수 중요도"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "image/png": {
       "height": 373,
       "width": 552
      },
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 변수 중요도 계산\n",
    "\n",
    "importances_values = pd.Series(ms_model.feature_importances_, index=x_valid.columns).sort_values(ascending=False)\n",
    "\n",
    "# 변수 중요도 시각화\n",
    "\n",
    "plt.figure(figsize=(8,6))\n",
    "plt.title('Feature Importances')\n",
    "sns.barplot(x=importances_values, y=importances_values.index)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 로지스틱 : 0.6122\n",
    "- 랜덤 포레스트 : 0.7747\n",
    "- SVM : 0.7483\n",
    "- XGBoost : 0.7761"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}