XGBoost学习

回归问题

from xgboost import XGBRegressor
model =XGBRegressor()
train=model.fit(X, y)#X为特征属性，y为标签

分类问题：

from xgboost import XGBClassifier
model =XGBClassifier()
train=model.fit(X, y)#X为特征属性，y为标签

代码

可以寻找最佳迭代次数，均方值误差，模型精确度

def modelfit(alg, dtrain, predictors, useTrainCV=True, cv_folds=5, early_stopping_rounds=50):
    if useTrainCV:
        xgb_param = alg.get_xgb_params()
        xgtrain = xgb.DMatrix(dtrain, label=predictors)
        # XGBoost有一个很有用的函数“cv”，这个函数可以在每一次迭代中使用交叉验证，并返回理想的决策树数量。
        cvresult = xgb.cv(xgb_param, xgtrain, num_boost_round=alg.get_params()['n_estimators'], nfold=cv_folds,
                          metrics='auc', early_stopping_rounds=early_stopping_rounds)
        t=np.arange(0,len(cvresult["train-auc-mean"]))
        alg.set_params(n_estimators=cvresult.shape[0])
    # Fit the algorithm on the data
    alg.fit(dtrain, predictors, eval_metric='auc')
    dtrain_predprob = alg.predict_proba(dtrain)[:, 1]
    # Print model report:
    print("\nModel Report")
    print("Accuracy : %.4g" % explained_variance_score(predictors, dtrain_predictions))
    print("AUC Score (Train): %f" % metrics.explained_variance_score(predictors, dtrain_predprob))

    feat_imp = pd.Series(alg.get_booster().get_fscore()).sort_values(ascending=True)
    print("feat_imp", "*" * 30)
    print("feature_name feature_importance_score")
    b=list(feat_imp.index)
    print(feat_imp.index)
    print(list(feat_imp.index))
    # print(len(list(feat_imp)))
    # print(len(feat_imp))
    return cvresult.shape[0]