LionelMassoulard
LionelMassoulard
next steps, make the tools available in auto-ml * create a parameter to say that the classification is ordinal (not sure we can that) * add 'composition' step that make...
I fixed everything ... if you can take a look
see https://github.com/societe-generale/aikit/pull/70
to finish to include latest functionnalities of pipeline
should wait for sklearn to be removed from experimental
Other usefull thing : see if LGBM can work with multi-target (maybe by creating a custom loss summing over the 2 target ?)