UKNN Implicit

This page analyzes the hyperparameter tuning results for the ItemKNN scorer in implicit-feedback mode.

Parameter Search Space

Parameter Type Distribution Values Selected
max_nbrs Integer Uniform 2 ≤ \(x\) ≤ 50 47
min_nbrs Integer Uniform 1 ≤ \(x\) ≤ 5 2
min_sim Float LogUniform 1e-06 ≤ \(x\) ≤ 0.1 1e-05

Final Result

Searching selected the following configuration:

{'max_nbrs': 47, 'min_nbrs': 2, 'min_sim': 1.0017972543183347e-05}

With these metrics:

{
    'RBP': 0.23404099424134026,
    'NDCG': 0.41976977802420784,
    'RecipRank': 0.4162145784771224,
    'TrainTask': 'ef35d789-466a-48e3-ac54-0e99446f4c9f',
    'TrainTime': 0.1816444438882172,
    'TrainCPU': 0.5365879999999997,
    'TestTask': '1a506ba0-0be7-4bb7-9518-390f8b1e45be',
    'TestTime': 1174.3037158311345,
    'TestCPU': 3043.050819,
    'timestamp': 1742787881,
    'checkpoint_dir_name': None,
    'done': True,
    'training_iteration': 1,
    'trial_id': '082be_00046',
    'date': '2025-03-24_03-44-41',
    'time_this_iter_s': 1177.607836484909,
    'time_total_s': 1177.607836484909,
    'pid': 3549206,
    'hostname': 'cci-p102',
    'node_ip': '10.246.250.62',
    'config': {'max_nbrs': 47, 'min_nbrs': 2, 'min_sim': 1.0017972543183347e-05},
    'time_since_restore': 1177.607836484909,
    'iterations_since_restore': 1,
    'experiment_tag': '46_max_nbrs=47,min_nbrs=2,min_sim=0.0000'
}

Parameter Behavior

Neighborhood Size

The neighborhood size is key argument to consider, let’s look at RMSE with repspect to that:

Filtering

There are two filtering parameters we also test — minimum neighbors and mininmum simialrity.