问题
From Liaw's classification and regression by RF paper, "The best way to determine how many trees are necessary is to compare predictions made by a forest to predictions made by a subset of forest"
I am wondering if there is a way to extract subtree for prediction with R's randomForest package. getTree seems to print out the structure. Any suggestion would be greatly appreciated.
回答1:
Try this one in randomForest, predict(rf, dat, predict.all=TRUE), you can get predictions from all the sub trees.
来源：https://stackoverflow.com/questions/21567893/extract-a-subset-of-tree-from-random-forest-model-for-prediction