预估器建模方式

1.Xgboost建模，sklearn评估
2.网格搜索交叉验证找最优参数
3.early-stop早停止
4.特征重要度
5.并行训练加速

#预估器建模方式：sklearn与XGboost配合使用
#xgboost建模，sklearn评估
import pickle
import xgboost as xgb

import numpy as np
from sklearn.model_selection import KFold, train_test_split, GridSearchCV
from sklearn.metrics import confusion_matrix, mean_squared_error
from sklearn.datasets import load_iris, load_digits, load_boston

rng = np.random.RandomState(31337)

#二分类：混淆矩阵
print("数字0和1的二分类问题")
digits = load_digits(2) #2表示类别的意思
y = digits['target']
X = digits['data']
#数据切分对象
kf = KFold(n_splits=2,shuffle=True,random_state=rng)#对数据进行切分
print("在2着数据上的交叉验证")
#2折交叉验证
for train_index,test_index in kf.split(X):
    xgb_model = xgb.XGBClassifier().fit(X[train_index],y[train_index])#初始化XGB分类器，拟合数据
    predictions = xgb_model.predict(X[test_index])#预估
    actuals = y[test_index]
    print("混淆矩阵")
    print(confusion_matrix(actuals,predictions))

#多分类，混淆矩阵
print('\nIris:多分类'