Is it possible to slim down a GBM model, in R? -

is possible remove data gbm fitted model? aware of keep.data option. however, i'd able keep data during processing slim down. also, if want keep single tree future predictions (let's after fitting 1000 trees, decided 500 trees works best in case, i'd keep tree. there safe way remove excess data fitted model without compromising ability predict new data?

i think helpful way remove excess trees suggest.

if run following on gbm example, trees make 80% of object size, , data 10%. deleting excess trees save lot of space.

library(pryr) bit <- unlist(lapply(gbm1,object_size)) round(bit/sum(bit),3) 

you might try compressing object, reduces it's size significantly.

memcompress(serialize(gbm1, null), "bzip2") unserialize(memdecompress(gbm1, type = "bzip2"))