ALS BiasedMF

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

Parameter Search Space

<frozen abc>:106: UserWarning: component class Placeholder does not define a config attribute type

Parameter	Type	Distribution	Values	Selected
embedding_size_exp	Integer	Uniform	3 ≤ \(x\) ≤ 10	9
regularization.user	Float	LogUniform	1e-05 ≤ \(x\) ≤ 1	0.0117
regularization.item	Float	LogUniform	1e-05 ≤ \(x\) ≤ 1	0.187
damping.user	Float	LogUniform	1e-12 ≤ \(x\) ≤ 100	40.1
damping.item	Float	LogUniform	1e-12 ≤ \(x\) ≤ 100	4.47e-09

Final Result

Searching selected the following configuration:

{
    'embedding_size_exp': 9,
    'regularization': {'user': 0.011747132123865267, 'item': 0.18724989438997966},
    'damping': {'user': 40.10579006914584, 'item': 4.467504166868086e-09},
    'epochs': 5
}

With these metrics:

{
    'RBP': 0.01945305503105208,
    'DCG': 9.186969523026955,
    'NDCG': 0.37136955902410673,
    'RecipRank': 0.057353423168151695,
    'Hit10': 0.14986824769433466,
    'RMSE': 0.7499156594276428,
    'max_epochs': 30,
    'epoch_train_s': 65.01545819593593,
    'epoch_measure_s': 24.105327347991988,
    'done': True,
    'training_iteration': 5,
    'trial_id': 'e7e07ed2',
    'date': '2025-09-29_23-09-20',
    'timestamp': 1759201760,
    'time_this_iter_s': 89.12855172157288,
    'time_total_s': 458.4700496196747,
    'pid': 3890089,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {
        'embedding_size_exp': 9,
        'regularization': {'user': 0.011747132123865267, 'item': 0.18724989438997966},
        'damping': {'user': 40.10579006914584, 'item': 4.467504166868086e-09},
        'epochs': 5
    },
    'time_since_restore': 458.4700496196747,
    'iterations_since_restore': 5
}

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

User Reg.

User Damp.

Item Reg.

Item Damp.

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?