Class: Rumale::EvaluationMeasure::MutualInformation

Inherits:
Object
  • Object
show all
Includes:
Base::Evaluator
Defined in:
lib/rumale/evaluation_measure/mutual_information.rb

Overview

MutualInformation is a class that calculates the mutual information.

Reference

  • N X. Vinh, J. Epps, and J. Bailey, “Information Theoretic Measures for Clusterings Comparison: Variants, Properties, Normalization and Correction for Chance,” J. Machine Learning Research, vol. 11, pp. 2837–1854, 2010.

Examples:

evaluator = Rumale::EvaluationMeasure::MutualInformation.new
puts evaluator.score(ground_truth, predicted)

Instance Method Summary collapse

Instance Method Details

#score(y_true, y_pred) ⇒ Float

Calculate mutual information

Parameters:

  • y_true (Numo::Int32) —

    (shape: [n_samples]) Ground truth labels.

  • y_pred (Numo::Int32) —

    (shape: [n_samples]) Predicted cluster labels.

Returns:

  • (Float) —

    Mutual information.



23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
 
# File 'lib/rumale/evaluation_measure/mutual_information.rb', line 23

def score(y_true, y_pred)
  check_label_array(y_true)
  check_label_array(y_pred)
  # initiazlie some variables.
  mutual_information = 0.0
  n_samples = y_pred.size
  class_ids = y_true.to_a.uniq
  cluster_ids = y_pred.to_a.uniq
  # calculate mutual information.
  cluster_ids.map do |k|
    pr_sample_ids = y_pred.eq(k).where.to_a
    n_pr_samples = pr_sample_ids.size
    class_ids.map do |j|
      tr_sample_ids = y_true.eq(j).where.to_a
      n_tr_samples = tr_sample_ids.size
      n_intr_samples = (pr_sample_ids & tr_sample_ids).size
      if n_intr_samples.positive?
        mutual_information +=
          n_intr_samples.fdiv(n_samples) * Math.log((n_samples * n_intr_samples).fdiv(n_pr_samples * n_tr_samples))
      end
    end
  end
  mutual_information
end