Ben Sadeghi

I'm open to both approaches for the new structs. It'd be good to experiment a bit and see which is more efficient, in terms of model training execution time and...

Thanks for trying out and benchmarking the package. I've been meaning to add support for DataFrames and model formula syntax, but just haven't gotten around to it. As for merging...

Yes, good catch, there is a discrepancy between the purity measures used in tree building and tree pruning. It would be nice to make ```prune_tree``` criterion agnostic, but I'm be...

Adult dataset, see test/classification folder. Try a fully grown tree, a forest of 10 trees and an adaboost of 10 iterations.

@Eight1911 The native API has been around for a while, and there are quite a few packages using it, see bottom of [this page](https://juliaobserver.com/packages/DecisionTree) in the "Used By" section. Adding...

@Eight1911 Perfect, thanks!

Now that you mention it, perhaps we should extend all hard-coded ```Float64``` types to subtypes of ```AbstractFloat```. This hasn't been an issue before, since```Float64``` is the default float regardless of...

I've added a couple tests covering ```Int32``` runs, for classification and regression, added adult and iris datasets to the test suite, and also updated the random and digits tests. Regarding...

I'm all for using a [0, 1] range for pruning purity. To maintain continuity on ```prune_tree```'s usage, we would need a ```purity_thresh``` of 1.0 to represent pure leaves (ie no...

I had a go at using a normalized entropy measure to map `pruning_thresh` to range [0,1], which worked, but its usability was still not intuitive. The same goes for using...