Fig. 6

Confusion matrices. Heatmap of confusion matrices from classification (random forest) using features extracted using BioCAT (a) compared with landmark and semi-landmark data (b). The data in (a) and (b) is shown together in (c) to facilitate comparison. Numbers represent percentage of correct classifications. lm_* represent the landmark/semi-landmark data. BioCAT features were mis-classified more consistently as some genotypes, e.g. mis-classification of mam mutants as Star (a). This pattern is not evident in the classification using the landmark data (b). Scale represents frequency of classification