使用train方法训练模型后,无法使用fit方法的bst.evals_result(),报错如下:
'Booster' object has no attribute 'evals_result'
采用以下方法:
immport xgboost as xgb
param={...}
xg_train = xgb.DMatrix(x_train, label=y_train)
xg_test = xgb.DMatrix(x_test, label=y_test)
evals_result = {}#记录训练集误差和验证集误差
bst=xgb.train(param,
xg_train,
evals=[(xg_train, 'Train'), (xg_test, 'Valid')],
num_boost_round=num_round,
evals_result=evals_result,
verbose_eval=True)
#输出训练集误差和验证集误差
print(evals_result)
绘制学习曲线:
train_loss=list(evals_result['Train'].values())[0]
valid_loss=list(evals_result['Valid'].values())[0]
x_scale=[i for i in range(len(train_loss))]
plt.figure(figsize=(10,10))
plt.title('loss')
plt.plot(x_scale,train_loss,label='train',color='r')
plt.plot(x_scale,valid_loss,label='valid',color='b')
plt.legend()
plt.show()
如下图所示: