daniel servén
daniel servén
we dont need to us `np.zeros(...)` to initialize an empty sparse matrix (since this would also defeat the purpose)
fixes https://github.com/dswah/pyGAM/issues/187 https://github.com/dswah/pyGAM/issues/76 fixes https://github.com/dswah/pyGAM/issues/124 write an example like pomegranate out of core: https://pomegranate.readthedocs.io/en/latest/ooc.html - [X] QR updating - [x] documentation - [x] all methods avoid using full model matrix...
it could be nice to add gamma to the gridsearch method. this way a user can manually choose how much to oversmooth
pg 164 in Wood introduces this concept. elaborated in appendix A.5
negative binomial is good for modeling counts where there are more zeros than expected by a poisson model http://dept.stat.lsa.umich.edu/~kshedden/Courses/Stat600/Notes/glm.pdf http://www.biostat.umn.edu/~dipankar/bmtry711.11/lecture_13.pdf
closes https://github.com/dswah/pyGAM/issues/196 - [x] add `decision_function` method to the `LogisticGAM` class for compatibility with sklearn's `OneVsRestClassifier` class - [x] test - [ ] refactor terms so that all argument processing...
We need to validate the data-dependent model parameters after validating the data! this fixes https://github.com/dswah/pyGAM/issues/245 and fixes https://github.com/dswah/pyGAM/issues/246 and fixes https://github.com/dswah/pyGAM/issues/232
for large search-spaces, gridsearch can easily exhaust all available memory before fitting a single model. This is because the method tries to pre-determine the entire search space in memory. we...
this will make possible to search a huge model spaces.