do you use xgboost? Maybe we should have a meetup hacknight to make a clojure interface to xgboost... https://xgboost.readthedocs.io/en/latest//get_started/index.html thoughts? @pandeiro
Good idea. We need more Clojure data science tools. Any idea of how long it might take?
If we assume it is already installed, perhaps it becomes an interop translation. Perhaps estimating the time might be the first step 😉
there is also the component that should translate a collection of clojure maps into libsvm txt format file, sparse matrix in CSR/CSC format, and dense matrix formats
actually, the format part may already exist...
https://github.com/r0man/svm-clj/blob/master/src/svm/core.clj
or here https://github.com/mattak/cl-svm
Sounds interesting, thanks @aaelony - I'm all for it but pretty inexperienced in this domain, myself.
The way I see it, it's win-win. There's (1) a (potential) data to libsvm format portion that reformats the data, (2) a Clojure to Java or Scala interop portion, and (3) the validation that it all works is the comparison to one/some of the examples in other languages. (4) is the freebie of learning a bit about boosting.