Thursday, June 7
Quite a work day! I already had the visualization code done for the first project, but all that it does is render shapes of variable size at variable locations (actually pretty much everything is variable). It still has to get those numbers from somewhere. I wrote a companion program in C++ to sift through the data and count the primary and secondary data relationships to plug into the visualizer. Even after that, though, it’s just an image. To turn it into a useful tool for a data scientist, it needs labeling. I added text by hand to say what each node represents and to label each node and edge with how many responses it represents. Here’s the result:
It’s important to note that this isn’t designed for public consumption. The use-case here is for researchers to explore a set of data to look for trends or surprising relationships. (For example, everyone who uses a space heater also wears extra clothing in the winter.) As such, this visualization needs to be easy to use, and not necessarily easy to learn.
I worked with Ilan a bit while evaluating different options for presentation methods and behind-the-scenes organization, so thank you for that!
Also someone pulled the fire alarm a few hours after we learned where to evacuate to. Suspicious…