Figure 1: Multi-Target Visual Debugging Workflow.(1) A user selects a model from a set of multi-target classifiers to inspect its performance. (2) Through filtering of parl ially correcl results and the visual investigation between the relationships of the n-target, the user can focus on a certain single-target. (3) By inspecting the filtered single-target confusion of multi-classes, a dri ll down onto the unde rlying time series classification data is possible that aims to give meaningful insights into the dass confusions.