CODEWITHSUNDEEP
pythonscikit-learnrandom-forest
user2766123
user2766123

Reputation: 115

Scikit-learn, random forests - How many samples does each tree contain?

In scikit-learn's RandomForestClassifier, there is no setting to specify how many samples each tree should be built from. That is, how big the subsets should be that are randomly pulled from the data to build each tree.

I'm having trouble finding how many samples scikit-learn pulls by default. Does anyone know?

Upvotes: 3

Views: 375

Answers (1)

Oleksi
Oleksi

Reputation: 13097

I believe RandomForestClassier will use the entire training set to build each tree. Typically building each tree involves selecting the features which have the most predictive power(the ones which create the largest 'split'), and having more data makes computing that more accurate.

Upvotes: 1

Related Questions