Ryan Bressler comments

Results 26 comments of


                                            Ryan Bressler

Document how to interpret importance.tsv

There is some documentation in the README: https://github.com/ryanbressler/CloudForest#importance Let me know if you have any specific questions.

panic: interface conversion: *CloudForest.AdaBoostTarget is not CloudForest.BoostingTarget: missing method Boost

Oh darn i broke that in a recent update, will try to find a chance to fix it soon.

Report specificity, sensitivity etc for binary classification with `-test`

I'm not sure how useful it is to have these reported by the utility (i mostly export the predictions and do my validation elsewhere using roc auc) but they could...

Output Class Vote Counts from applyforest, maybe add quantile or raw vote output for regression?

Done for categorical variables. Will be harder for numerical variables as a running mean is used.

Slow performance splitting on large categorical variables.

I was unclear but it is actually both, a user provided me with a data set with 200k+ samples mostly high cardinality features. I attached a pprof screen shot though...

organize or extract some useful utils

The code has definitely outgrown all being in one package but I haven't had time to reorganize it. I'm happy to provide feedback on proposals and accept pull requests though....