I've added a QuantileRegression class to the ML Toolkit (using GradientBoostingRegressor from sklearn), and am working on finding a good way to fit a rolling prediction method as well (similar to the MAD showcase). Is there a way to contribute back to the module?
Hi ianroddis,
At present we don't have the machinery in place to incorporate external updates to the app, but I'm glad to hear about and would be happy to see the results of your work. Posting things on Answers isn't a bad idea!