Michael McDaniel
Comparing Checkout Times + Volumes of Dewey Classes

I have attempted to create a tool that may be used to compare the relative times and volumes at which particular pairs of dewey classes are checked out. For this project we were asked to increase the complexity of earlier visualizations. I therefore chose to use an entire months worth of checkout data, thus resulting in an engrossing and complex visual experience.

By placing a mark at each minute of the day when an item of a given dewy class was checked out I was hoping to capture a qualitative sense of the number of items checked out, the volume, and the time of day when the majority of checkout activity occurred. Juxtaposing two Dewey classes allows for direct qualitative comparisons of respective class checkout patterns. Note that one pixel on the Y axis equals one minute.

Layout

The layout of this visualization was constrained by the sheer number of data being visualized. With upwards to 50,000 data points, I had to chose dimensions and an orientation that would capture all the data on a 1024 x 768 resolution projection screen. My first draft saw the hours on the X axis with days on the Y axis. The dimensions required for this orientation left little room to add text and labels. Therefore, in my final visualization hours are found on the Y axis with days on the X axis. This orientation allows for a more effective utilization space.

The graph itself is positioned 60 pixels from the right and left and 90 and 45 pixels from the top and bottom, respectively. Had space allowed, the graph would have been placed 120 pixels from the top and 60 pixels from the bottom.

Color

Using ColorBrewer and suggestions from the in-class critique, I settled on the color palate below. The blue/pruple and orange are distinct but have roughly equal weight making them effective indicators of the different Dewey categories. The color coated Dewey numbers and names located above the graph serve as descriptors as well as color keys.

I chose two shades of gray in order to emphasize some design elements while relaxing others. I chose to include the dark gray horizontal lines to break the data up into hour-long segments. Days are distinguished by light gray columns on alternating days. Weeks are indicated by gray shaded boxes surrounding Sundays at the top and dates at the bottom.

See Iterations below for some example of other color palates I tried.

Type + Text

I have used two fonts in this visualization, Lucida Sans for lettered elements and Gill Sans for numeric elements. I chose Lucida Sans because it is clean and simple yet compact and Gill Sans as it is favored by Tufte for numbering, and I find it works well with Lucida.

It seems helpful to have both the day of the week as well as the date present so I have placed the abbreviated day name at the top of the graph and the date at the bottom.

To balance the hour starting times on the left, and to give a stronger visual cue for time of day, I placed hour ending times on the right. The hour dividers are also extended out beyond the graph boundaries on each side.

Problems

Some of the visual elements, namely the hour dividers, week indicators and large amount of text, become more prominent than I would like and add to the already busy visualization. However, without these cues some patterns in the data might be overlooked.

The title text on the top right conflicts with the Dewey text on the left. Without it though, I don't believe the user is afforded enough information to easily understand what the visualization is communicating.

Find the evolution of this project here.