Bias Predictor

This page analyzes the behavior and hyperparameter search for the Bias scoring model on ML10M.

Parameter Search Space

Parameter	Type	Distribution	Values	Selected
damping.user	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	32.6
damping.item	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	6.35

Final Result

Searching selected the following configuration:

{'damping': {'user': 32.5852499977984, 'item': 6.347456543685538}}

With these metrics:

{
    'RBP': 0.09411695856950328,
    'LogRBP': 1.4310229397354988,
    'NDCG': 0.39266434570885084,
    'RecipRank': 0.265141224557562,
    'RMSE': 0.7833646285287515,
    'TrainTask': 'e4c792e8-32df-400b-88bc-524ba8243083',
    'TrainTime': 2.0552472840063274,
    'TrainCPU': 2.0516389999999998,
    'TestTask': 'd5452743-e29d-4959-8e52-ee3150886b84',
    'TestTime': 6.947036410972942,
    'TestCPU': 6.956925999999999,
    'timestamp': 1746452749,
    'checkpoint_dir_name': None,
    'done': True,
    'training_iteration': 1,
    'trial_id': '09ab8_00072',
    'date': '2025-05-05_09-45-49',
    'time_this_iter_s': 12.727388858795166,
    'time_total_s': 12.727388858795166,
    'pid': 4038830,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {'damping': {'user': 32.5852499977984, 'item': 6.347456543685538}},
    'time_since_restore': 12.727388858795166,
    'iterations_since_restore': 1,
    'experiment_tag': '72_item=6.3475,user=32.5852'
}

Metric Response

How does RMSE change with each setting independently?

User Damping

Item Damping

Best Configurations

Since this is an explicit-feedback rating prediction model, our primary search criteria is RMSE. The configuration with the best RMSE is:

	config.damping.user	config.damping.item	RBP	RMSE
Method
Random	32.58525	6.347457	0.094117	0.783365

If we instead searched for RBP, we would select:

	config.damping.user	config.damping.item	RBP	RMSE
Method
Random	5.125235	96.629343	0.105091	0.790671

Search Geometry

What is the geometry of the search space?