Source code for immuneML.ml_metrics.ClusteringMetric

import numpy as np
import sklearn

INTERNAL_EVAL_METRICS = ['calinski_harabasz_score', 'davies_bouldin_score', 'silhouette_score']
EXTERNAL_EVAL_METRICS = ['rand_score', 'adjusted_rand_score', 'adjusted_mutual_info_score', 'completeness_score',
                         'fowlkes_mallows_score', 'homogeneity_score', 'mutual_info_score',
                         'normalized_mutual_info_score', 'v_measure_score']




def compute(metric: str, features, predictions, distance_metric: str = None):
    metric_fn = getattr(sklearn.metrics, metric)
    if metric == 'silhouette_score' and distance_metric == 'precomputed':
        return metric_fn(features, predictions, metric='precomputed')
    elif distance_metric == 'precomputed':
        return np.nan
    else:
        return metric_fn(features, predictions)





def is_internal(metric: str):
    return metric in INTERNAL_EVAL_METRICS





def is_external(metric: str):
    return metric in EXTERNAL_EVAL_METRICS





def is_valid_metric(metric: str):
    return metric in INTERNAL_EVAL_METRICS or metric in EXTERNAL_EVAL_METRICS




def get_search_criterion(metric):
    assert is_valid_metric(metric), f"Metric '{metric}' is not recognized as a valid clustering metric."

    if metric in ['adjusted_mutual_info_score', 'adjusted_rand_score', 'calinski_harabasz_score', 'completeness_score',
                  'fowlkes_mallows_score', 'homogeneity_score', 'mutual_info_score', 'normalized_mutual_info_score',
                  'rand_score', 'silhouette_score', 'v_measure_score']:
        return max
    elif metric in ['davies_bouldin_score']:
        return min
    else:
        raise ValueError(f"Metric '{metric}' is not recognized as a valid clustering metric.")