search

liudragonfly / GBDT

A simple GBDT in Python
353 stars 193 forks source link

为什么没有subsample #11

Open Kiris-tingna opened 5 years ago

Kiris-tingna commented 5 years ago

我记得GBDT里有shrinkage和subsample的过程啊 代码里只有列采样

zhaochangming commented 5 years ago

代码里有shrinkage与subsample这个啊，如下： shrinkage： for node in leaf_nodes: for id in node.get_idset(): f[id][label] += learn_ratenode.get_predict_value() subsample： subset = sample(subset, int(len(subset)self.sample_rate)) 还是说我理解错了？

liudragonfly commented 5 years ago

@zhaochangming 你是对的