Interactive Pattern Analysis of Multiple T-Maze Data

Abstract

The Multiple T-Maze study is one of the standard methods used in ethology and behaviourism. In this paper we extend the current state of the art in analysis of Multiple T-Maze data for animal cohorts. We focus on pattern finding within animals' paths. We introduce the Sequence View which makes it possible to quickly spot patterns and to search for specific sub-paths in animal paths. Further, we also evaluate four different metrics for string comparison and two widely used embeddings to support interactive clustering. All views are fully integrated in a coordinated multiple views system and support active brushing. This research represents a step towards (semi)-automatic clustering for Multiple T-Maze cohort data, which will significantly improve the Multiple T-Maze data analysis.