Beyond horizon graphs: space efficient time series visualization with composite visual mapping

Restricted screen space is a limit to visualization of time series regardless of the medium. To address this challenge, we introduce new space-efficient visual designs for time series based on an approach similar to the well established Horizon Graph, namely visual mapping. In this approach, each data attribute is decomposed into two components and then each component is mapped onto a separate visual channel. Our visual designs consist in different combinations ofgeometric and optical visual channels. We compare our propositions with Horizon Graph across different chart heights and we measure accuracy and speed of users in a discrimination and estimation task. Our results show that although Horizon Graph perform best at larger chart heights, our propositions demonstrate same levels of accuracy in small chart heights. Moreover, at least one of our propositions has significant advantage over Horizon Graphs in terms of speed. Based on our findings, we propose design guidelines for using composite visual mapping and combinations of optical visual channels in limited vertical screen resolution settings.