是否有可能在XGBoost算法中指定叶片节点的最小重量分数？

发布于 2025-02-01 04:00:30 字数 383 浏览 3 评论 0原文

我试图实现一种简单的XGBoost算法，并希望指定我的样品的最小重量分数，应在每个叶子节点中实现。到目前为止，我只找到了参数min_child_weight，该参数指定了节点中的最小样本数。我想知道我是否有一种方法可以在此处结合每个样品的重量，因为不是每个样本的重量= 1？

如果有人可以在这里提供帮助，我会非常高兴！谢谢:)

下面您可以找到我的代码：

xg_r = xg.XGBRegressor(objective="reg:squarederror", n_estimators = 10, max_depth = 5, min_child_weight = 500000)
xg_r.fit(X_train, y_train)
xg.plot_tree(xg_r)

原文

I tried to implement a simple XGBoost algorithm and would like to specify a minimum weight fraction of my samples, that should be fulfilled in every leaf node. So far I only found the parameter min_child_weight, which specifies the minimum number of samples in a node. I would like to know if there is a way of how I can incorporate the weight of each sample here, since not each sample has weight=1?

I would be super glad if someone can help here! Thank you :)

Below you can find my code:

xg_r = xg.XGBRegressor(objective="reg:squarederror", n_estimators = 10, max_depth = 5, min_child_weight = 500000)
xg_r.fit(X_train, y_train)
xg.plot_tree(xg_r)

分享到QQ

分享到微博