Like the title says. Would it be possible to add an option, say "independent = True" to OneHotEncoder that would return a 1 of k-1 encoding instead of a 1 of k encoding. This would be very useful to me when I am encoding categorical variables since the 1 of k encoding adds an extra (non-independent) degree of freedom to the model. It would also be nice if I could specify which category to keep as the baseline.

Something like:

X = np.array([12,24,36]).reshape(-1,1)  
OneHotEncoder(sparse=False, independent=True, baseline=24).fit_transform(X)  
Output: array([[ 1., 0.],
       [ 0., 0.],
       [ 0., 1.]])