[R] Warning “Unknown or uninitialised column: `ntree`.” when trying to pass hyperparameters to a learner with package mlr

Agnes g2g @gne@g2g @end|ng |rom hotm@||@com
Tue May 4 22:23:41 CEST 2021


​Dear all,
I posted this question on stats exchange and got the message that it was off-topic.https://stats.stackexchange.com/questions/522745/warning-unknown-or-uninitialised-column-ntree-when-trying-to-pass-hyperpar
They suggested to use this email list instead.
So that's why I am posting it here. I have read the instructions, but if I have overlooked something that said this question does not belong on this mailing list, I apologise.

Thanks in advance for reading this message!

Kind regards,
Agnes Wattel
problem

I want to do a grid search with different hyperparameters that are provided by a self-made grid. But when I am running the code I get a warning: "Unknown or uninitialised column: ntree."

create grid

gs <- list(ntree = c(50,100,150,200,300,500,550),mtry = c(1,2,3,4,5)) %>% cross_df()


select hyperparameters

  for (h in 1:nrow(gs)) {
    ntree_entry <- gs$ntree[h]
    mtry_entry <- gs$mtry[h]
    (...)
    lrn <- makeMultilabelClassifierChainsWrapper(lrn, order = NULL)
    lrn <- setPredictType(lrn,"prob")
    task <- makeMultilabelTask(data = data[train_lines,], target = label_bact)
    mod <- mlr::train(lrn,task)
    }


On the place of (...) I have tried two different ways to get the hyperparameters into the learner:

option 1: pass hyperparameters to the learner

lrn <- makeLearner("classif.randomForest",ntree = ntree_entry,mtry = mtry_entry)


option 2: pass hyperparameters to the learner

lrn <- makeLearner("classif.randomForest")
lrn <- setHyperPars(lrn,par.vals = list(ntree = ntree_entry,mtry = mtry_entry))


check

The console tells me that the hyperparameters are in the learner:

>lrn
Learner classif.randomForest from package randomForest
Type: classif
Name: Random Forest; Short name: rf
Class: classif.randomForest
Properties: twoclass,multiclass,numerics,factors,ordered,prob,class.weights,oobpreds,featimp
Predict-Type: response
Hyperparameters: ntree=50,mtry=1


ntree and mtry are hyperparameters that are known to the makeLearner, because they are both in getParamSet:

> getParamSet("classif.randomForest")

                     Type  len   Def   Constr Req Tunable Trafo
ntree             integer    -   500 1 to Inf   -    TRUE     -
mtry              integer    -     - 1 to Inf   -    TRUE     -
replace           logical    -  TRUE        -   -    TRUE     -
classwt     numericvector <NA>     - 0 to Inf   -    TRUE     -
cutoff      numericvector <NA>     -   0 to 1   -    TRUE     -
strata            untyped    -     -        -   -   FALSE     -
sampsize    integervector <NA>     - 1 to Inf   -    TRUE     -
nodesize          integer    -     1 1 to Inf   -    TRUE     -
maxnodes          integer    -     - 1 to Inf   -    TRUE     -
importance        logical    - FALSE        -   -    TRUE     -
localImp          logical    - FALSE        -   -    TRUE     -
proximity         logical    - FALSE        -   -   FALSE     -
oob.prox          logical    -     -        -   Y   FALSE     -
norm.votes        logical    -  TRUE        -   -   FALSE     -
do.trace          logical    - FALSE        -   -   FALSE     -
keep.forest       logical    -  TRUE        -   -   FALSE     -
keep.inbag        logical    - FALSE        -   -   FALSE     -


Question

Why do I get the message "Unknown or uninitialised column: ntree"? Although it is only a warning, not an error, I am afraid that it will only use the first line of the grid that I created. The sizes of the models seem to confirm that. They all have the same size. I hope that someone can help me with this problem. Thanks in advance!


	[[alternative HTML version deleted]]



More information about the R-help mailing list