ALS BiasedMF

This page analyzes the hyperparameter tuning results for biased matrix factorization with ALS.

Parameter Search Space

Parameter	Type	Distribution	Values	Selected
embedding_size	Integer	LogUniform	4 ≤ \(x\) ≤ 512	327
regularization.user	Float	LogUniform	1e-05 ≤ \(x\) ≤ 1	0.65
regularization.item	Float	LogUniform	1e-05 ≤ \(x\) ≤ 1	0.00269
damping.user	Float	LogUniform	1e-12 ≤ \(x\) ≤ 100	66.9
damping.item	Float	LogUniform	1e-12 ≤ \(x\) ≤ 100	1.75e-10

Final Result

Searching selected the following configuration:

{
    'embedding_size': 327,
    'regularization': {'user': 0.649746030842825, 'item': 0.002690798332663406},
    'damping': {'user': 66.89592055271686, 'item': 1.746296335934768e-10},
    'epochs': 6
}

With these metrics:

{
    'RBP': 0.05363133913048115,
    'NDCG': 0.3568942374137382,
    'RecipRank': 0.09746205750688007,
    'RMSE': 0.7654107316018957,
    'TrainTask': '8978a200-058c-4186-bc9f-b9dcc620926e',
    'TrainTime': None,
    'TrainCPU': None,
    'max_epochs': 30,
    'done': True,
    'training_iteration': 6,
    'trial_id': 'a81e89dc',
    'date': '2025-04-30_23-49-53',
    'timestamp': 1746071393,
    'time_this_iter_s': 100.89020133018494,
    'time_total_s': 690.7189295291901,
    'pid': 400825,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {
        'embedding_size': 327,
        'regularization': {'user': 0.649746030842825, 'item': 0.002690798332663406},
        'damping': {'user': 66.89592055271686, 'item': 1.746296335934768e-10},
        'epochs': 6
    },
    'time_since_restore': 690.7189295291901,
    'iterations_since_restore': 6
}

Parameter Analysis

Embedding Size

The embedding size is the hyperparameter that most affects the model’s fundamental logic, so let’s look at performance as a fufnction of it:

Learning Parameters

Iteration Completion

How many iterations, on average, did we complete?

How did the metric progress in the best result?

How did the metric progress in the longest results?