Performance Summary

This is a summary view of model performance on the ML25M data set.

Top-N Recommendation Accuracy

Rating Prediction Accuracy

Resource Consumption

Time

Memory

Power

Leaderboard Table

This provides quick numeric access to the model results by mean metric. Note, however, that simple means can be misleading!

Top-N Recommendation

		RBP	NDCG	RecipRank
model	variant
flexmf-bpr	optuna-best	0.215	0.149	0.396
flexmf-logistic	optuna-best	0.210	0.146	0.390
als-implicit	optuna-best	0.189	0.139	0.360
flexmf-warp	optuna-best	0.187	0.130	0.358
iknn-implicit	optuna-best	0.165	0.113	0.325
iknn-implicit	default	0.154	0.111	0.304
popular	default	0.139	0.092	0.305
als-implicit	default	0.132	0.113	0.254
flexmf-warp	default	0.124	0.090	0.257
flexmf-logistic	default	0.107	0.070	0.243
bias	optuna-best	0.098	0.072	0.266
flexmf-explicit	optuna-best	0.095	0.071	0.222
flexmf-bpr	default	0.095	0.067	0.238
als-biased	default	0.062	0.053	0.155
iknn-explicit	default	0.006	0.014	0.023
iknn-explicit	optuna-best	0.005	0.014	0.021
flexmf-explicit	default	0.003	0.006	0.015
als-biased	optuna-best	0.002	0.002	0.007
bias	default	0.001	0.001	0.008

Rating Prediction

		RMSE	MAE
model	variant
als-biased	optuna-best	0.792	0.635
iknn-explicit	optuna-best	0.794	0.637
iknn-explicit	default	0.795	0.637
als-biased	default	0.796	0.640
flexmf-explicit	optuna-best	0.801	0.643
flexmf-explicit	default	0.811	0.652
bias	optuna-best	0.814	0.659
bias	default	0.824	0.667