jcrudy
commented
6 years ago @florakarniav In scikit-learn terminology, what you want is a classifier, while Earth is a regressor. It's possible to create a classifier from one or more Earth models, but it might be easier to just use an existing classifier such as RandomForest. See http://scikit-learn.org/stable/modules/ensemble.html#random-forests. If you do want to form a classifier from Earth models, the simplest way is to create a Pipeline with two steps, the first being an Earth model and the second a LogisticRegression model. It's also possible to use gradient boosting to produce a classifier from multiple Earth models, but the code to do that is not well-tested or packaged yet.
Hello, This is more of a question than an issue. I would like to know if it is possible to configure MARS to make predictions of distinct values.
For example if my Y takes values in 2,4,8, is there any way to make y_hat produce results consisting of these values only?
Thank you in advance