亚麻Applied Scientist电面

uniwander 发表于 2019/04/05 11:53:22
thanks for sharing

补充内容 (2019-4-5 20:28):
n*K*logn?
1. initiate K means, 2. iterate thru all point...

为啥logn？我查了一下有两个版本：
1. (n^(d*k+1))*logn
n-number of points to cluster, d-dimensions
2. ndk generally

感谢楼主，祝福

请问楼主是general hire, 还是ML 方向的组呢？谢谢

onehorselee 发表于 2019/04/12 01:52:25
请问楼主是general hire, 还是ML 方向的组呢？谢谢

TRMS组

about the RF. I think if you need to remove the features depend on the number of features(computation cost).  For the decision Tree, there is no Need to remove correlated features. If you check the algorithm of the decision tree. It will automatically ignore the correlated features if its' correlated features already been chosen ad node before.