자동차보험 가입여부 예측분류하기(KAGGLE사례)

데이터 설명 : 자동차 보험 가입 예측 (종속변수 Response: 1 : 가입 , 0 :미가입)
x_train: https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/x_train.csv

y_train: https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/y_train.csv
x_test: https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/x_test.csv
x_label(평가용) : https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/y_test.csv
데이터 출처 :https://www.kaggle.com/anmolkumar/health-insurance-cross-sell-prediction(참고, 데이터 수정)

 

 

반드시 캐글 예시 사례 봐보자!!(boosting/EDA)

https://www.kaggle.com/code/yashvi/vehicle-insurance-eda-and-boosting-models

Vehicle Insurance EDA and boosting models

from sklearn.model_selection import RandomizedSearchCV
from sklearn.model_selection import cross_val_score
from sklearn.ensemble import RandomForestClassifier
from catboost import CatBoostClassifier
from scipy.stats import randint
import pickle
import xgboost as xgb
import lightgbm as lgb
from sklearn.metrics import accuracy_score
# import packages for hyperparameters tuning
from hyperopt import STATUS_OK, Trials, fmin, hp, tpe
from sklearn.model_selection import train_test_split, RandomizedSearchCV, StratifiedKFold, KFold, GridSearchCV
from sklearn.metrics import f1_score, roc_auc_score,accuracy_score,confusion_matrix, precision_recall_curve, auc, roc_curve, recall_score, classification_report 

 

 

import pandas as pd
#데이터 로드
x_train = pd.read_csv("https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/x_train.csv")
y_train = pd.read_csv("https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/y_train.csv")
x_test= pd.read_csv("https://raw.githubusercontent.com/Datamanim/datarepo/main/insurance/x_test.csv")


display(x_train.head())
display(y_train.head())


# print(x_train.info())
# print(x_train.nunique())
# print(x_train.isnull().sum())



# id는 제가 잘못넣은 컬럼입니다.
drop_col = ['ID','id']
x_train_drop = x_train.drop(columns = drop_col)
x_test_drop = x_test.drop(columns = drop_col)

# import sklearn
# print(sklearn.__all__)
# import sklearn.model_selection
# print(dir(sklearn.model_selection))

from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
x_train_dummies = pd.get_dummies(x_train_drop)
y = y_train['Response']


x_test_dummies = pd.get_dummies(x_test_drop)
# train과 컬럼 순서 동일하게 하기 (더미화 하면서 순서대로 정렬을 이미 하기 때문에 오류가 난다면 해당 컬럼이 누락된것)
x_test_dummies = x_test_dummies[x_train_dummies.columns]
# print(help(train_test_split))



X_train, X_validation, Y_train, Y_validation = train_test_split(x_train_dummies, y, test_size=0.33, random_state=42,stratify=y)
rf = RandomForestClassifier(random_state =23)
rf.fit(X_train,Y_train)

# import sklearn.metrics
# print(dir(sklearn.metrics))

from sklearn.metrics import accuracy_score , f1_score, recall_score, roc_auc_score ,precision_score

#model_score
predict_train_label = rf.predict(X_train)
predict_train_proba = rf.predict_proba(X_train)[:,1]

predict_validation_label = rf.predict(X_validation)
predict_validation_prob = rf.predict_proba(X_validation)[:,1]


# 문제에서 묻는 것에 따라 모델 성능 확인하기
# 정확도 (accuracy) , f1_score , recall , precision -> model.predict로 결과뽑기
# auc , 확률이라는 표현있으면 model.predict_proba로 결과뽑고 첫번째 행의 값을 가져오기 model.predict_proba()[:,1]
print('train accuracy :', accuracy_score(Y_train,predict_train_label))
print('validation accuracy :', accuracy_score(Y_validation,predict_validation_label))
print('\n')
print('train f1_score :', f1_score(Y_train,predict_train_label))
print('validation f1_score :', f1_score(Y_validation,predict_validation_label))
print('\n')
print('train recall_score :', recall_score(Y_train,predict_train_label))
print('validation recall_score :', recall_score(Y_validation,predict_validation_label))
print('\n')
print('train precision_score :', precision_score(Y_train,predict_train_label))
print('validation precision_score :', precision_score(Y_validation,predict_validation_label))
print('\n')
print('train auc :', roc_auc_score(Y_train,predict_train_proba))
print('validation auc :', roc_auc_score(Y_validation,predict_validation_prob))


# test데이터 마찬가지 위와 같은 방식
predict_test_label = rf.predict(x_test_dummies)
predict_test_proba = rf.predict_proba(x_test_dummies)[:,1]


# accuracy, f1_score, recall, precision 
#pd.DataFrame({'ID': x_test.ID, 'Response': predict_test_label}).to_csv('003000000.csv', index=False)

# auc, 확률
#pd.DataFrame({'ID': x_test.ID, 'Response': predict_test_proba}).to_csv('003000000.csv', index=False)