In a SAS Viya gradient boosting model (gbtreetrain CAS action), a data scientist sets SUBSAMPLERATE=0.8 and COLSAMPLEBYTREE=0.7. What do these parameters together create?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because SUBSAMPLERATE=0.8 means each boosting tree is trained on 80% of the training observations randomly sampled without replacement (row subsampling), and COLSAMPLEBYTREE=0.7 means 70% of variables are randomly sampled for each tree (column subsampling). Together these implement stochastic gradient boosting, increasing diversity among trees and reducing variance.
Full explanation below image
Full Explanation
B is correct because SUBSAMPLERATE=0.8 means each boosting tree is trained on 80% of the training observations randomly sampled without replacement (row subsampling), and COLSAMPLEBYTREE=0.7 means 70% of variables are randomly sampled for each tree (column subsampling). Together these implement stochastic gradient boosting, increasing diversity among trees and reducing variance. A transposes the parameters. C confuses these with depth/leaf parameters. D incorrectly associates subsampling with L1 regularization.