Bias Predictor

This page analyzes the behavior and hyperparameter search for the Bias scoring model on ML1M.

Parameter Search Space

Parameter	Type	Distribution	Values	Selected
damping.user	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	5.3
damping.item	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	2.78

Final Result

Searching selected the following configuration:

{'damping': {'user': 5.301282302114991, 'item': 2.778205189402948}}

With these metrics:

{
    'RBP': 0.013039721392061216,
    'LogRBP': -0.5455151185795941,
    'NDCG': 0.1762810501898602,
    'RecipRank': 0.04528146129336854,
    'RMSE': 0.8574191222016682,
    'TrainTask': '6b87d7cc-1a8e-4b77-bf94-bcfafe82c5dc',
    'TrainTime': 0.24627943901577964,
    'TrainCPU': 0.2386389999999997,
    'TestTask': '4785d24c-3e13-4c0b-99af-30a5a61825e7',
    'TestTime': 1.1866091410047375,
    'TestCPU': 1.1818449999999998,
    'timestamp': 1746443408,
    'checkpoint_dir_name': None,
    'done': True,
    'training_iteration': 1,
    'trial_id': '7203e814',
    'date': '2025-05-05_07-10-08',
    'time_this_iter_s': 2.187699794769287,
    'time_total_s': 2.187699794769287,
    'pid': 3827677,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {'damping': {'user': 5.301282302114991, 'item': 2.778205189402948}},
    'time_since_restore': 2.187699794769287,
    'iterations_since_restore': 1,
    'experiment_tag': '54_item=2.7782,user=5.3013'
}

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	5.301282	2.778205	0.01304	0.857419

If we instead searched for RBP, we would select:

	config.damping.user	config.damping.item	RBP	RMSE
Method
Random	5.247436	95.387517	0.021799	0.867328

Search Geometry

What is the geometry of the search space?