Task-driven evaluation of aggregation in time series visualization

Abstract

Many visualization tasks require the viewer to make judgments about aggregate properties of data. Recent work has shown that viewers can perform such tasks effectively, for example to efficiently compare the maximums or means over ranges of data. However, this work also shows that such effectiveness depends on the designs of the displays. In this paper, we explore this relationship between aggregation task and visualization design to provide guidance on matching tasks with designs. We combine prior results from perceptual science and graphical perception to suggest a set of design variables that influence performance on various aggregate comparison tasks. We describe how choices in these variables can lead to designs that are matched to particular tasks. We use these variables to assess a set of eight different designs, predicting how they will support a set of six aggregate time series comparison tasks. A crowd-sourced evaluation confirms these predictions. These results not only provide evidence for how the specific visualizations support various tasks, but also suggest using the identified design variables as a tool for designing visualizations well suited for various types of tasks.

Investigating Time Series Visualisations to Improve the User Experience

“…This was functionally identical to the maxima task, except that "highest" was changed to "lowest". Despite their similarities, prior work [4,32] suggests that there are differences in the performance of these two tasks and that different visualisations may be more suitable. 3.…”

“…Following previous graphical perception studies [4,10,14,16,21], we used synthetic time series data in order to have greater control on the data values and their corresponding visual representation. We generated 96 distinct time series datasets, one for each experiment condition.…”

“…Each dataset had 112 data points (1 per day) and spanned over a period of 16 weeks. Based on our pilot testing and the data density of datasets used by similar studies [4,10,14], we estimated that this scale of data would be substantial enough to achieve the appropriate level of difficulty for the different tasks of our study.…”

“…The aforementioned trend of focusing more on visual representation than on interaction is also predominant in graphical perception of time series visualisations, which is a well-studied area in visual analytics. A common practice in quantitative graphical perception studies of time series visualisations [4,10,14,16,21] is to conduct them in a static setting, i.e., not allowing the users to interact with the time series visualisations, thereby limiting our knowledge of the user experience. In addition, the overall coverage of different visual representations of time series data is lacking.…”

“…In addition, the overall coverage of different visual representations of time series data is lacking. Several previous studies [4,10,14,16,21] have compared the effectiveness of time series visualisations within and across the categories of positional (e.g., Figure 1(a&b)) and colour visual encodings (e.g., Figure 1(c&d)). However, to the best of our knowledge, there has been no study that compares the effectiveness of time series visualisations that use area visual encoding (e.g., Figure 1(e&f)) with their positional and colour counterparts.…”