xgboost使用

定义绘制以训练样本数为横坐标的学习曲线的函数

完整代码

import numpy as np
from sklearn.datasets import load_boston
from sklearn.model_selection import train_test_split,learning_curve,KFold
from xgboost import XGBRegressor as xgbr
import matplotlib.pyplot as plt

data=load_boston().data
target=load_boston().target
X_train,X_test,y_train,y_test=train_test_split(data,target,test_size=0.3,random_state=420)

def plot_learning_curve(estimator,title,x,y,
                        ax=None,
                        ylim=None,
                        cv=None,
                        n_jobs=None):
    train_sizes,train_scores,test_scores=learning_curve(estimator,x,y,
                                                        shuffle=True,
                                                        cv=cv,
                                                        n_jobs=n_jobs,
                                                        # random_state=420,
                                                        )
    if ax==None:
        ax=plt.gca()
    ax.set_title(title)
    if ylim is not None:
        ax.set_ylim(*ylim)
    ax.set_xlabel('Training Example')
    ax.set_ylabel('score')
    ax.grid()
    #axis=1:按行计算的均值
    ax.plot(train_sizes,np.mean(train_scores,axis=1),'o-',c="r",label="Train__score")
    ax.plot(train_sizes,np.mean(test_scores,axis=1),'o-',c="g",label="Train__score")
    ax.legend(loc="best")
    return ax
cv = KFold(n_splits=5, shuffle = True, random_state=42)

plot_learning_curve(xgbr(n_estimators=100,random_state=420),
                        "XGB",X_train,y_train,ax=None,cv=cv)

plt.show()

细化学习曲线，找出最佳n_estimators

n_est=range(10,1010,100)
rs=[]
def plot_cvs(model):
    for i in n_est:
        reg=model(n_estimators=i,random_state=420)
        rs.append(cross_val_score(reg,X_train,y_train).mean())

plot_cvs(xgbr)
plt.figure(figsize=(3,4))
plt.title('XGB')
print('n_est:',n_est)
print('rs:',rs)
plt.plot(n_est,rs,c="red",label="XGB")
plt.legend()
plt.show()