A three step process to design visualisations for GeoTemporal analysis

Abstract

Given vehicle tracking data, loyalty and credit card logs of employees from a fictitious company, GAStech, participants of VAST 2014 mini challenge 2 were tasked to extract the common daily routine of employees and identify any suspicious activities that may be present in the data. In this paper, we reflect on our analysis procedure focusing on each step of the process that contributed to problem solving. Accordingly, we describe the features incorporated into our software at each stage of the process and justify the design decisions that were made. Inspired by DiBiase's approach to visual analysis [1], our procedure consists of three stages (Fig. 1). With off-the-shelf software, such as R and QGIS, we conducted exploratory data analysis [2] to generate a diverse range of insights. The insights were evaluated based on their relevance to the given task. They were used to formulate hypotheses and data task abstraction [3] resulting in a set of complementary tools comprising of an origin-destination map, a timeline and a flow diagram that we developed in processing. These tools were not designed to function as an integrated software package but were treated as rapid prototypes that would afford more flexibility in the design and development cycle [4].