Jason Rudy

Right, I keep them in for two reasons: 1. Consistency with scikit-learn, which keeps them in. 2. Elimination of cython dependency/sensitivity to cython version I do not believe the C...

@Fish-Soup If you want to quantify volatility, you can probably come up with something based on the `predict_deriv` method of the fitted `Earth` model. It returns the gradient of your...

You can actually do this right now using a combination of py-earth and scikit-learn. Just pass `enable_pruning=False` to the `Earth` constructor, then put the `Earth` model in a `Pipeline` with...

I don't know the advantages or disadvantages of using elasticnet vs the regular pruning method. You'll have to experiment. I would be curious to know the results. Regarding stability of...

@Fish-Soup I'm not super familiar with scikit-learn's elastic net, but if it's expecting normalized data then you might try putting a `StandardScaler` in your pipeline between the `Earth` and `ElasticNet`...

@Fish-Soup I'm not sure why the Elastic Net isn't behaving as expected. Perhaps the scikit-learn people can give you advice on that. In general, you can perform the steps of...

@Fish-Soup One of the out columns from an `Earth` model's transform method is a constant, representing the intercept term. If that's what's causing problems, you could add a stage to...

@Fish-Soup It would be possible to do something like that in some limited form, although it isn't currently implemented. It would depend a lot on the type of function and...

@Fish-Soup I just realized when responding to your other post that I forgot to respond to this one. It seems like you can fit f2 separately from MARS (using some...

For 1, you can just use sample weights based on lag time. However, I don't know by what principle you would calculate the exact weights. Perhaps you could figure out...