Design of Visual Analytics Systems

Tay, Jebb, and Woo (2017) highlight that the following four issues need to be carefully considered in the design of visual analytics systems: (a) identification (isolating and highlighting relevant data and patterns, and the relevant scale of analysis), (b) integration (combining different data sources and different models to reveal new insights), (c) immediacy (streaming, real-time, and time-sensitive data, identify important dynamic changes over time using both incoming and historical data), and (d) interactivity (to select, switch, swap, and combine different data types on visual interfaces to inductively uncover and identify new patterns). The field of visual analytics expands on previous work to assist researchers, analysts, and decision-makers in their use of data for effective discovery, monitoring, analysis, and decision-making (Song et al., 2020). These aspects enhance the utility and usability of visual analytics. Utility refers to the ability of the system to support users in completing the required tasks, and usability describes the ease of using the system, according to the analyst perception, in completing the same required tasks (Ellis & Dix, 2006). To ensure utility and usability of a system, visual analytics researchers usually adopt the user-centred design paradigm, and work closely with stakeholders (e.g., directors, managers, coaches, trainers, analysts) at various stages of design and development. User-centred design usually entails identifying the context of use, specifying requirements, and creating design solutions. Creating solutions is typically an iterative process in which multiple design ideas are presented to users (via sketches or actual implementations of the system), feedback is sought, and intermediate assessments are conducted, leading to refining the design and presenting the system again for more feedback (Song et al., 2020). After a system implementation is completed, researchers conduct final evaluations through various protocols (Ellis & Dix, 2006) to report on the usability and utility of the system for the particular objective and users.

Research in computational methods, visualization, and cognitive science can help advance visual analytics by finding solutions that leverage both machine computational power and human intelligence. This is important not only to visual analytics but also to computational methods because automated methods are also marked by the choices (and potentially biases) that humans introduce when designing algorithms, sampling data, and interpreting the results (Karimzadeh et al., 2020).

< Prev   CONTENTS   Source   Next >