Скачать или смотреть Understand & Implement Normalized Mutual Information (NMI) in Python (normalized_mutual_info_score)

Normalized Mutual Information (NMI) provides a quantitative assessment of a clustering algorithm's ability to identify meaningful and distinct groups within the data

Given the knowledge of the ground truth class assignments `labels_true` and our clustering algorithm assignments of the same samples `labels_pred`, the mutual Information measures the agreement of the two assignments

creating a joint probability matrix,
which represents the joint occurrences of unique values from both arrays
marginal probabilities for each array
are calculated by summing the rows and columns of the joint probability matrix
The entropies of each array are calculated using the marginal probabilities
The mutual information is computed by summing over all possible pairs of values
Normalized Mutual Information (NMI)

#clustering #datascience #machinelearning