Fig. 4
From: SMILE: systems metabolomics using interpretable learning and evolution
![Fig. 4](http://media.springernature.com/full/springer-static/image/art%3A10.1186%2Fs12859-021-04209-1/MediaObjects/12859_2021_4209_Fig4_HTML.png)
Model testing accuracy in relation to the number of effective features. For all models using a given number of effective features, a violin plot shows the distribution of their testing accuracy. The correlation between testing accuracy and the number of effective features is neither strong nor significant