Bias Predictor

This page analyzes the behavior and hyperparameter search for the Bias scoring model on ML20M.

Parameter Search Space

Parameter	Type	Distribution	Values	Selected
damping.user	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	57.6
damping.item	Float	LogUniform	0.1 ≤ \(x\) ≤ 100	5.17

Final Result

Searching selected the following configuration:

{'damping': {'user': 57.602503822663856, 'item': 5.1749118450847815}}

With these metrics:

{
    'RBP': 0.09695452501991363,
    'NDCG': 0.3640569455949519,
    'RecipRank': 0.2721170677074266,
    'RMSE': 0.798532027317719,
    'TrainTask': 'ea235cd2-36d6-4e5a-b66a-d638f2a8ed83',
    'TrainTime': 3.999668174998078,
    'TrainCPU': 4.010676999999999,
    'TestTask': '0ce7e703-e286-4625-acff-295091aa4611',
    'TestTime': 10.028997720997722,
    'TestCPU': 10.038702999999998,
    'timestamp': 1745423638,
    'checkpoint_dir_name': None,
    'done': True,
    'training_iteration': 1,
    'trial_id': 'cd7e2_00045',
    'date': '2025-04-23_11-53-58',
    'time_this_iter_s': 19.166555404663086,
    'time_total_s': 19.166555404663086,
    'pid': 101119,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {'damping': {'user': 57.602503822663856, 'item': 5.1749118450847815}},
    'time_since_restore': 19.166555404663086,
    'iterations_since_restore': 1,
    'experiment_tag': '45_item=5.1749,user=57.6025'
}

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	57.602504	5.174912	0.096955	0.798532

If we instead searched for RBP, we would select:

	config.damping.user	config.damping.item	RBP	RMSE
Method
Random	5.125235	96.629343	0.098681	0.806339

Search Geometry

What is the geometry of the search space?