Fig. 5From: SMILE: systems metabolomics using interpretable learning and evolutionIndividual feature importance ranking. A The overall feature importance ranking using 1000 evolved models. B A feature occurrence graph generated on SMILE’s web interface using 176 filtered models with a higher than 80% testing accuracy and using two effective features. The x-axis shows feature index and y-axis shows feature occurrence frequency. The corresponding feature name will be shown with mouse hoverBack to article page