(draft)
Where is the randomness of Random Forest.
The first is to select the randomness of the training samples, and then put them back
The second is the randomness of selecting tree node features. Selecting the optimal nsub sub-features among the n features will increase the degree of summation of the tree to the sample, but will improve the generalization ability. Cross-validation can be used to determine the appropriate nsub.
https://www.cnblogs.com/pinard/p/6156009.html