Bias Predictor

This page analyzes the behavior and hyperparameter search for the Bias scoring model on ML25M.

Parameter Search Space

Parameter	Type	Distribution	Values	Selected
damping.user	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	19.7
damping.item	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	5.4

Final Result

Searching selected the following configuration:

{'damping': {'user': 19.673032636525292, 'item': 5.404067637457645}}

With these metrics:

{
    'RBP': 0.0983122894186686,
    'LogRBP': 1.474633729646957,
    'NDCG': 0.36509723046970244,
    'RecipRank': 0.26672304573739114,
    'RMSE': 0.81378388230229,
    'TrainTask': '681731f5-1a1d-4f1a-b0ab-f95b5e0dcad1',
    'TrainTime': 4.609336641966365,
    'TrainCPU': 4.612283000000001,
    'TestTask': '6c2c68bf-7cb0-4808-8601-473cba767bb0',
    'TestTime': 12.06055010703858,
    'TestCPU': 12.087274,
    'timestamp': 1746483583,
    'checkpoint_dir_name': None,
    'done': True,
    'training_iteration': 1,
    'trial_id': 'ea4aa2a4',
    'date': '2025-05-05_18-19-43',
    'time_this_iter_s': 26.59047269821167,
    'time_total_s': 26.59047269821167,
    'pid': 297565,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {'damping': {'user': 19.673032636525292, 'item': 5.404067637457645}},
    'time_since_restore': 26.59047269821167,
    'iterations_since_restore': 1,
    'experiment_tag': '29_item=5.4041,user=19.6730'
}

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	19.673033	5.404068	0.098312	0.813784

If we instead searched for RBP, we would select:

	config.damping.user	config.damping.item	RBP	RMSE
Method
Random	0.409147	66.235385	0.100502	0.821918

Search Geometry

What is the geometry of the search space?