What insights can data visualization techniques generate?

Visualizing information is a way to communicate scientific results efficiently and in an appealing way. This is particularly important if scientists want to communicate their results not only to their academic peers but also to the general public or policy makers.In fact, researchers are increasingly expected to come up with appealing, sophisticated graphs if they want to publish their work in highly ranked journals.

In fact, researchers are increasingly expected to come up with appealing, sophisticated graphs if they want to publish their work in highly ranked journals.

On Friday, the 28th of February, Katerina Vrotsou from the Visualization Center in Norrköping (Norrköping Campus of the Linköping University) gave a presentation on visualizing information on events at the Institute for Futures Studies. She used daily activity diary data from 463 people, also including socio-demographic information. The activities were labeled using seven categories like “care for self” which included activities like sleeping and food intake, “gainful employment/school” or “reflection/recreation” which included activities like reading or watching the TV.

Katerina developed a toolbox for visualization analysis, VISUAL-TimePAcTS, that allows various kinds of visualizations of event-sequence data, such as bar charts, where the full day and night of each individual is visualized using different colors for the seven activity categories. Fading out other activities in the bar it’s possible to highlight single activities. The activity diagrams of men and women can be compared and the bars can be re-arranged by age for instance.

Furthermore, Katerina implemented a sequential mining algorithm to identify sequences of activities, that is, several single activities that would usually appear together. Cooking, eating and washing dishes, for example, are identified as a sequence. Other visualization techniques that Katerina presented were interactive exploration graphs and clustering diagrams that identify groups of individuals that behave similarly.

While applying these visualization techniques to everyday activities data might not generate any surprising new insights (yes, we first cook, then eat and then clean up, and yes women are cooking and cleaning more often than men), these techniques may be quite useful in other contexts and with other data. Sequences of events might be less obvious in other contexts than the everyday life that we are all familiar with. For instance, it can be very helpful to get a first notion of what patterns are hidden in the data if one is interested in analyzing Big data. Good visualization techniques may allow scientist to get their message across more straightforward. Therefore, learning from and collaborating with visualization experts is definitely an asset.

by Viktoria Spaiser