def get_stats()

in src/rime/dataset/base.py [0:0]

• 34 lines of code
• 3 McCabe index (conditional complexity)

    def get_stats(self):
        if "TEST_START_TIME" in self.user_in_test and "_hist_ts" in self.user_in_test:
            avg_hist_span = self.user_in_test[  # test users with finite history
                (self.user_in_test["TEST_START_TIME"] < np.inf) &
                (self.user_in_test["_hist_ts"].apply(len) > 0)
            ].apply(
                lambda x: x["TEST_START_TIME"] - x["_hist_ts"][0], axis=1
            ).mean()
        else:
            avg_hist_span = float("nan")

        return {
            'user_df': {
                '# test users': len(self.user_in_test),
                '# train users': len(self.training_data.user_df),
                'avg hist len': self.user_in_test['_hist_len'].mean(),
                'avg hist span': avg_hist_span,
                'horizon': self.horizon,
                'avg target items': self.target_csr.sum(axis=1).mean(),
            },
            'item_df': {
                '# test items': len(self.item_in_test),
                '# train items': len(self.training_data.item_df),
                'avg hist len': self.item_in_test['_hist_len'].mean(),
                'avg target users': self.target_csr.sum(axis=0).mean(),
            },
            'event_df': {
                '# train events': len(self.training_data.event_df),
                '# test events': self.target_csr.sum(),
                'default_user_rec_top_c': self.default_user_rec_top_c,
                'default_item_rec_top_k': self.default_item_rec_top_k,
                "user_ppl_baseline": self.user_ppl_baseline,
                "item_ppl_baseline": self.item_ppl_baseline,
            },
        }