Closed roban closed 9 years ago
This looks correct to me, I had a bunch of style notes.
Thanks, I'll clean this up and test some more.
I'm working through the style fixes, but in the mean time I wrote some code to visualize the difference between a tree and a pruned version of the tree.
Here's an example:
Red nodes are the one's pruned from the tree. Green nodes are the replacement leaves.
@avibryant: thanks for all the tips. Really helpful as I get a handle on idiomatic scala. I'm going to do some testing with larger models on hadoop before I request another review.
This LGTM except that:
BTW an example of what I mean by hashing properly on the id is in Samplers: https://github.com/stripe/brushfire/blob/master/brushfire-core/src/main/scala/com/stripe/brushfire/Samplers.scala#L12
Should I worry about the Travis error?
As long as it builds, no, I think I broke travis earlier (and since we don't have any tests, it doesn't seem high priority to fix :)
Apart from my one minor comment, LGTM
Very much WIP at the moment, but it compiles, runs, and seems to to do something reasonable.
I added the digits dataset to have another example, and created
example/iris-prune
andexample/digits-prune
scripts (to run afterexample/iris
orexample/digits
).Advice welcome, @avibryant.