如何用python实现贝叶斯网络优化lightgbm超参数？

可以参考这个ppt

讲的就是用贝叶斯方法来调参

def krrcv(
		alpha,
		degree,
		coef0,
		):
		
	params['alpha'] = max(alpha, 0)
	params['degree'] = max(degree, 0)
	params['alpha'] = max(alpha, 0)
	data=train_df
	target=y_train_df
	val=cross_val_score(KernelRidge(params),data,target,scoring='r2',cv=5).mean()
	return val

krrBO=BayesianOptimization(krrcv,{
				        'alpha':(0,1),
					'degree':(0,4),
				        'coef0':(0,5)
					})
					

这是我目前用贝叶斯网络优化 KRR的参数，但是总是出错，比如

TypeError: ufunc 'add' output (typecode 'O') could not be coerced to provided ou
tput parameter (typecode 'd') according to the casting rule ''same_kind''

网上参考的代码太少了，希望能够得到帮助！