机构名称：
¥ 5.0
●Breiman（2001）首先提出了随机森林算法，但基于1995年的Tim Kan Ho●RF采用了两种集合技术：首先是训练样本，以种植基于不同培训训练数据的树木森林。第二个是特征空间的子采样。●如果我选择变量的子集（例如x1, x3, x7) to create a split in a node of a decision tree, and another subset (x2, x4, x5, x7) to create a different one, there will be events that get classified in a different way by the two nodes ● Often there is a dominant variables that is used to decide the split, offsetting the power of the subdominant ones.rf通过减少不同树的相关性来避免该问题
监督与无监督的学习
主要关键词
相关文件推荐
