Fig. 7From: Discovering feature relevancy and dependency by kernel-guided probabilistic model-building evolutionRelevancy and dependency discovery in the Hepatitis dataset using Kiedra (see the discussion of these results in the text). a Description of the variables in the dataset. b A heatmap of the discovered relevancy factors; each row correspond to the factors comprising the best solution of one repetition of the experiment. Relevant (white) and irrelevant (shaded) variables where chosen with a cutoff value of 0.5 on the average factors over all repetitions (only repetitions obtaining a classification accuracy greater than 80% were considered). c A graph of aggregated dependencies obtained over all repetitions. Opacity indicates the frequency of repetition a dependency was found. d The minimum-spanning-tree on the aggregated graph, suggesting the final estimated dependencies. Relevancy is also shown (irrelevant variables and associated dependencies are shaded)Back to article page