【Task3(2天)】 模型构建

• 用逻辑回归、svm和决策树；随机森林和XGBoost进行模型构建，评分方式任意，如准确率等。（不需要考虑模型调参）时间：2天

数据集下载

说明：这份数据集是金融数据（非原始数据，已经处理过了），我们要做的是预测贷款用户是否会逾期。表格中 "status" 是结果标签：0表示未逾期，1表示逾期。

1.数据处理

#这里用根据IV选择出来的特征
features = list(iv_result[iv_result['iv_value']>0.4]['feature'])
X = df[features]
y = df['status']
from sklearn.model_selection import train_test_split   #数据切分
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.3,random_state=2019)
from sklearn.preprocessing import StandardScaler  #标准化
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.fit_transform(X_test)

2.模型导入，模型评估

二分类问题，我用准确率，查准率，召回率，f1指标来评估模型

#导入模型
from sklearn.linear_model import  LogisticRegression
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from xgboost import XGBClassifier
log_clf = LogisticRegression()
svc_clf = SVC()
tree_clf = DecisionTreeClassifier()
forest_clf = RandomForestClassifier()
xgb_clf = XGBClassifier()
models = {'log_clf':log_clf,'svc_clf':svc_clf,'tree_clf':tree_clf,'forest_clf':forest_clf,'xgb_clf':xgb_clf}

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

def metrics(models,X_train_scaled,X_test_scaled,y_train,y_test):
    results = pd.DataFrame(columns=['recall_score','precision_score','f1_score','accuracy_score'])
    for model in models:
        name = str(model)
        result = []
        model = models[model]
        model.fit(X_train_scaled,y_train)
        y_pre = model.predict(X_test_scaled)
        result.append(recall_score(y_pre,y_test))
        result.append(precision_score(y_pre,y_test))
        result.append(f1_score(y_pre,y_test))
        result.append(accuracy_score(y_pre,y_test))
        results.loc[name] = result
    return results

metrics(models,X_train_scaled,X_test_scaled,y_train,y_test)

xGBoost效果要好点，但总体看效果不是很好！！