Improve Random Forest note

[martinapugliese]

improve the note, see comments here

[martinapugliese]

• bagging reduces then variance keeping the bias constant [1]
• bagging needed not to correlate trees (work on same data) [1]
• only$$ \sqrt p $$ chosen at each node split, preventing trees from using the same features so they are decorrelated [1]
• OOB same as CV leave one out

[martinapugliese]

• [ ] on sparse datasets
• [ ] params to tune
• [ ] pruning of trees
• [ ] number of features : The reason for "randomness" in the algorithm is in the addition of, at each split in the learning process, a random subset of the features (feature bagging) %TODO mmm, is this it? This way, if a few features are strong predictors of the response variable, these will be selected by many trees, causing those to the correlated. The literature recommends, for a set of $p$ features, to use $\sqrt{p}$ of them for classification and $p/3$ for regression %TODO cite and explain