python_交叉验证

python_交叉验证

from sklearn.ensemble import RandomForestClassifier,GradientBoostingClassifier
from sklearn.cross_validation import train_test_split,cross_val_score,StratifiedKFold
from sklearn.metrics import accuracy_score,roc_auc_score,confusion_matrix,roc_curve,auc,precision_recall_curve,roc_auc_score
from sklearn import grid_search

columns = data.columns.drop(['last_etl_acg_dt','cust_isn','bel_org'])[0:10]
temp = data[columns]

# 交叉验证
RF = RandomForestClassifier()
aucs = cross_val_score(estimator=RF,X=temp.iloc[0:100000,:],y = [int(x>20) for x in data.cust_age.iloc[0:100000]],cv = 3,scoring='roc_auc')
print(aucs)

GBDT = GradientBoostingClassifier()
aucs = cross_val_score(estimator=GBDT,X=temp.iloc[0:100000,:],y = [int(x>20) for x in data.cust_age.iloc[0:100000]],cv = 3,scoring='roc_auc')
print(aucs)