Should I standardize my features before or after applying Splines?
More specifically, I am running the following code to transform my features:
transformed_x = dmatrix("bs(Data, df=6, degree=3, include_intercept=False)-1",
{"Data":Data}, return_type='dataframe')
which results in a cubic spline. Then I estimate my betas using something like:
GroupLassoRegressor(group_ids=Group_ids, alpha=0.5).fit(transformed_x, y_train)
My question is not so much about the implementation, but rather whether I should standardize my features (remove mean and scale to unit variance) before transforming them?
