Hi, Scikit-learn owners and contributors,

I am wondering whether scikit-learn want to implement the quantile regression forest: http://www.jmlr.org/papers/volume7/meinshausen06a/meinshausen06a.pdf.

It is a quite highly cited paper, and already has an R package. https://cran.r-project.org/web/packages/quantregForest/quantregForest.pdf

This method has been widely used in various quantile regression problems.

From implementation perspective, it is also a natural extension of random forest, given scikit-learn already has a good random forest implementation. Most of the computation is performed with random forest base method.

In addition, R's extra-tree package also has quantile regression functionality, which is implemented very similarly as quantile regression forest. So if scikit-learn could implement quantile regression forest, it would be an relatively easy task to add it to extra-tree algorithm as well.

Please let me know if it is possible, Thanks.