Input of fit() and return value of predict_proba() method #152
Description
I try to have FMClassification class integrated with scikit-learn classes (e.g. sklearn.metrics, sklearn.pipe ). However the fit() method permits only {-1, 1} for the binary labels and predict_proba() method doesn't return the regular scikit-learn-like format, (n_samples, n_classes) array, but returns (n_samples, ) array. Now I'm using a simple and imperfect wrapper class on my own, but I want to know is there any reason so, and if you have a plan to change?