machine learning - Aggregating labels in GradientBoostingRegression -
i trying understand scikit-learn's gradient boosting regression algorithm. followed source code , have understanding of iterative construction of trees based on chosen loss function. couldn't figure out answer how take average of labels underlying estimators when invoke predict()
.
i followed function call down this line. here, scale
holds learning_rate
, if not provided, default 0.1. so, if use 500 trees, don't understand why multiplying each of 500 different labels, given sample, 0.1.
if direct me published paper explains in depth, appreciated.
Comments
Post a Comment