Old versus New: How to effectively display small and big data sets

One of the main take-away messages I received from this week’s reading is that goals dictate how data should be displayed. For example, the primary goal of statisticians is to report findings. This is accomplished by using well established methods that are applied to a wide range of applications. Additionally, ‘simple’ graphics are the rule. That is, statisticians do not use overly flashy colors, complex 3-D graphics, or other unnecessary features (e.g., animation). On the other hand, the primary goal of Infovis designers is to attract attention to the displayed data by using colorful graphs, videos, animation, etc.

As a researcher I have to agree with Gelman and Unwin (2013) that the most important objective of data visualization should be accuracy, precision, and understand ability. However, I do not believe that data visualization needs to be limited to a few simple graphs. I think there is tremendous potential to depict data that captivates the audience, whomever the audience may be. Additionally, in the era of “big” data traditional data display methods may not be applicable or sufficient. Consequently, new data visualization techniques should be developed and supported by both statisticians and Infovis designers. In turn, this would help ensure that big data is displayed accurately and attention worthy.

Big Data Visualization from my own Research:

I have used word frequencies displays (Wordle) in two studies developed by visual analytics researchers in Tempe. Figure 1 shows a screen shot of the Sentiment Analysis page from a movie interface. The sentiment displays tweets about a given movie made two weeks prior. Additionally, the Sentiment Analysis page was developed using 4,795,176 movie-related tweets for 214 movies. One of the issues we had is that participants wanted more than words. That is, they wanted numbers associated with the tweeted words. However, we have not been able to effectively modify the interface in that it displays both, what is being said and how many times. Lastly, although the sentiment page attracts interests and attention it does not provide a detailed account of the data (e.g., how many tweets were collected to develop the sentiment for each movie). Again, reinforcing the challenge of accomplishing both goals–to grab attention, and at the same time, to accurately depict data.

References

Gelman, A., & Unwin, A. (2013). Infovis and statistical graphics: different goals, different looks. Journal of Computational and Graphical Statistics, 22(1), 2–28.