HCDE210 Process Blog 6: Data Visualization

Tableau UI view of the Geographic Distribution of Burglaries in Seattle, WA in January 2016. Sorted by type.

Before this task, the only information visualization experience I had was largely done for science and statistics — MS Excel and RStudio, respectively. In this, I had not been exposed to multi-dimensional data visualization techniques in any capacity and Tableau was the first program I had used to approach such a task. During the studio session, we were guided through the basics of using the Tableau software suite, including the process of importing different types of data, filtering it, and using different types of visualization methods. This hands-on practice allowed us to get accustomed to the Tableau UI/UX, its strengths and limitations, and understand what kinds of data we could visualize. From this, we could start our sprint; and I chose to visualize some statistics and patterns in the context of burglaries in Seattle during January 2016. Three visualizations were produced; the first being a map of the burglary locations, the second, day-based heat map (highlight table) of incidences, and the third being a discrete area chart of incidence by hour; each of which attempted to illustrate the data as effectively and simply as possible. This set of visualizations was then uploaded onto Tableau Public for public viewing.

The problems I encountered were predomiantly based on the amount of data availiable; for many categories, there was not enough data to infer trends. Practically, that is a good thing (not an extreme amount of crime in Seattle) but in terms of identiying trends or hotspots, it was not so practical (less crime = less data on said crime, tough to come to conclusions). Other problems included figuring out how to do certain tasks on Tableau (e.g. filtering certain data, multimensionality), these of which were solved via reading discussions on the Tableau forums. Some things that could be done differently include becoming more comfortable with the functionality of the program being used before trying to create effective visualizations.

Final Data Visualization on Burglaries in Seattle. Left: Geographic distribution. Top right: Highlight table for burglaries by day. Bottom right: Discrete area chart of burglaries by hour.

Wildcard question: 
How do you feel about the Tableau software suite?
Before this task, I had only really made (quantitative) data visualizations in two dimensions (X vx Y) in MS Excel and a little bit in RStudio; and in my experience that was already a lot of work — getting trendlines to appear, getting colors to match, switching the X and Y axes, all of these was often annoying to do and in many cases quite esoteric and requiring a full understanding of the software. Tableau surprised me by automatically processing all the data in a form that was extremely easy — load up the data, then drag-and-drop the data you want visualized into the different axes bins or parameter settings and everything was automatically processed. Visualization types (graphs, maps, etc) were automatically recommended and chosen based on the data loaded; which makes the process of effectively visualizing relevant data easier.

Extending these practices into other fields:
Since the advent of computers, health informatics has continued to grow as a larger and larger part medicine, espeically in the preventative field. By collecting and processing both new and existing data on potential health factors such as environmental factors, genetic predispositions, and lifestyle choices, doctors and other health-related professionals ask questions about their patients’ lifestyle or genetic disposition — and have the quantitative data to back it up. They can then make better-informed choices on the health of their patients. These types of applications are vast and can be intensely personal — looking at trends in cell mutation, looking at geographically-linked disease incidence, comparing parasite incidence rates, identifying trends in personal health and much more. For example, in consumer-oriented fitness trackers like Fitbit and Jawbone, by tracking the trends in the users’ sleep patterns, exercise patterns, etc we can already start to make sure we get enough exercise and sleep, improve focus, and more — thus enhancing the human experience and condition. If we imagine taking these types of data (and possibly more in-depth e.g. heart rate, blood pressure) and applying it to a medicine-based context, we can then apply the same concepts to preventative or personalized medicine, in which we can stay healthy in a more effective manner.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.