multinb.py

from sklearn.naive_bayes import _BaseNB

class MultiNB(_BaseNB):
    def __init__(self, models_dict):
        self.models_dict = models_dict

    def fit(self, X, y, **kwargs):
        for column, model in self.models_dict.items():
            model.fit(X[:, column], y, **kwargs)

        self.fitted_ = True

    def _joint_log_likelihood(self, X):
        sample_model = next(iter(self.models_dict.values()))

        jil = np.zeros((X.shape[0], sample_model.class_count_.shape[0]))
        jil += sample_model.class_log_prior_

        for column, model in self.models_dict.items():
            jil += model._joint_log_likelihood(X[:, column])
            jil -= model.class_log_prior_

        return jil