Figure 3
From: Identification of subject-specific responses to footwear during running
Post-hoc evaluation of the support vector machine model performance as a function of individual (A) subjects and (B) footwear conditions. Performance evaluation of the models trained for the three employed classification tasks: subject x footwear (in blue on the left), subject (in yellow in the middle), and footwear (in orange on the right). The task-specific zero-rule baseline values are 0.8% (subject x footwear classification), 3.3% (subject classification), and 25.0% (footwear classification). The prediction accuracy is shown as violin plots with median (solid line), mean (white dot), interquartile range from Q1 to Q3 (grey bar), and individual values (coloured dots) of the prediction accuracy for individual (A) subjects and (B) footwear conditions. The numbers in the violin plots highlight subject IDs that are discussed in sections "Performance evaluation", "Subject classification (across footwear conditions)", and "Footwear classification (across subjects)". The figure was created using the MATLAB code provided by Bechtold et al. (2022)37.
