r - RandomForest proximity between training and test -
using randomforest
, want create low-level projection of instance proximities, produced mdsplot()
. however, not want training proximities, proximities between all instances (i.e. training and test). how can this?
i use xtest
argument randomforest()
, when do
print("testprox") print(rfmodel$test$prox) print("trainprox") print(rfmodel$prox)
i get
[1] "testprox" [1] 119 239 [1] "trainprox" [1] 120 120
so distance test all calculcated (119x239), not train all (120x239). instead, train train (120x120). how can full matrix all all (239x239)?
to missing piece, copy appropriate submatrix of test all (i.e. test train, 119x120) , use transpose plug in make full matrix. wondering if there easier way?
Comments
Post a Comment