Changing the metric normalizer is currently not supported, passing through average_method to compute_clustering_metrics provides access to other normalizers
def compute_clustering_metrics(..., average_method="max"):
# average_method = 'max' is *Normalized Information Distance (NID)*
# Follows metric property and range is [0, 1] (Vinh et al. 2010)
...
nmi = normalized_mutual_info_score(
labels_true=labels.true,
labels_pred=labels.pred,
average_method=average_method, # choices: min, geometric, arithmetic, max
)
https://github.com/KwanLab/Autometa/blob/350ff103ff21dfc6fab4d5e09b6e858e60b74cd6/autometa/validation/benchmark.py#L149-L153
Changing the metric normalizer is currently not supported, passing through
average_method
tocompute_clustering_metrics
provides access to other normalizersaverage_method
choices are listed in github.com/scikit-learn/sklearn/metrics/cluster/_supervised.py#L76-L89