search

bips-hb / blockForest

Random Forests for Blocks of Clinical and Omics Covariate Data
7 stars 4 forks source link

predict bug #10

Open jgiemza opened 4 years ago

jgiemza commented 4 years ago

Hello,

I can't run predict function on my current data.

blockfor_cv <- blockfor(data_tr_cv, y_tr_cv, num.trees = 2000, replace = FALSE, probability = FALSE, blocks=blocks, nsets = 300, num.trees.pre = 1500, splitrule="extratrees", importance = "impurity", block.method = "BlockForest", num.threads=10, always.select.block=0)

testpred <- predict(blockfor_cv$forest, data=data_te_cv, block.method=BlockForest, num.threads=10)

gives error:Error in[.data.frame(data, , forest$independent.variable.names, drop = FALSE) : undefined columns selected

Colnames of data_tr_cv are the same as in data_te_cv.

I have looked at your code and I guess the error corresponds to line 128 in predict.R. My response is binary. It should not be there I guess because I don't specify data as a formula.

I would be glad if you could help me.

mnwright commented 4 years ago

Could you give a reproducible example? I cannot reproduce your issue.