At the Edges of Time
A new way to add visual context to time-based charts.
Every day there is countless amounts of data being collected with time as the primary dimension, either continuously or in discrete intervals. A common way to visualize time-series data is to use time as the horizontal axis so that when it is read from right to left the viewer can see the measured variable rise or fall with relation to time. A primary limitation of time-based charts is that to display data that covers a large period of time, there is a tradeoff between detail in the data or a limit in the range of time covered in the visualization.
To combat these limitations, a new technique to visualize time-series data was developed by Morrow et al. and presented in a recent IEEE VIS paper. They refer to their creation as periphery plots; see below for an illustration:
A periphery plot may appear overwhelming at first but it is actually quite simple to read. The timeline display at the top shows the period of time that is being visualized below. The set of visualizations below the timeline are referred to as tracks, and one timeline can relate to any number of tracks for the applicable time period. The darker grey area marked as focus will be the period of time that has the expanded detail while the two lighter grey areas marked as context provide the range that will be represented in the periphery plots. The areas from the focus and control zones are applied to the same time frame on every track in the visualization.
The purpose of a periphery plot is to provide the broader context to the area of focus while having the context areas take up less physical space. This is accomplished by applying summarization techniques to the data that will be represented in the periphery plots, thus causing it to be less holistically informative but preserving some aspects of the data that allow it to be effective when viewed in conjunction with the fully detailed data displayed in the focus zone. Morrow et al. defined the types of summation in their paper as follows: time-value-axis-preserving which retains both the point in time that data is collected as well as the value associated with that data point, time-axis-preserving which retains the point in the that data is collected but not the value associated with the data, value-axis-preserving which represents the values of the data points but not the time they were collected, and no-axis-preserving which retains neither the time of collection or the value of the data. Examples of the types of preservation models are given in the following image:
Another key feature of the model developed by Morrow et al. is its potential for interactivity. The model allows for both zooming and panning which can provide even greater focus on certain areas of the data. Additionally, a user can change the bounds of the context and focus areas utilizing what is referred to as “multi-brush zone control.” One focus zone can also have any number of context zones created alongside it to allow for examinations of trends over larger periods of time.
The concept of periphery plots was developed to assist in visualizing health data for healthcare providers to be able to examine trends in patients over time. The first image showcases the versatility of the technique by using it to visualize climate data from Seattle. Periphery plots for future visualizations can allow for previously unseen trends to be more easily recognized in time-series data while reducing the amount of space required to display such data in a time based chart.
References
- Morrow, B., Manz, T., Chung, A. E., Gehlenborg, N., & Gotz, D. (2019). Periphery plots for contextualizing heterogeneous time-based charts. Proceedings of the IEEE Visualization Conference. https://doi.org/10.1109/visual.2019.8933582
