I’m lumping these together since my Tuesday post ended up pretty short.
Tuesday, June 5
We started with the “Craft of Research” class, where we addressed the reasons behind literature reviews and the questions behind our research. We talked about the ISU Institutional Review Board (and IRBs in general).
We attended a lunch lecture about biofuels and globalization by Mark Wright. He discussed the history of biofuels, world-wide energy sources, and predictions for the future. We asked about his work here at ISU and learned about the ongoing energy and biofuel projects and ISU’s leading position in the global biofuel research space.
Next, we (DataViz) had a meeting with our professor (Dr. Dorneich) and graduate student (Jacklin) about our lit review progress and our next mini task: make some visualizations to help researchers understand the results of a survey on household energy use. This will take some careful thinking, since we’re coming up with something new. We spent the rest of the day working on that, getting our pictures taken, blogging, eating, and sleeping.
Wednesday, June 6
Another C++ class in the morning. It gave me an opportunity to help people and some extra time to wake up.
We (DataViz) had a short meeting with Jacklin where we clarified some objectives and got feedback on a visualization we’ve been brainstorming on the whiteboard. After that, I started programming a way to create that visualization.
We (REU) left for lunch and made a stop at the administrative building to pick up our first paychecks. After making it back to the lab, I blogged and worked some more on the DataViz code.
To briefly summarize what we’ve been doing, I’ll give an example. Let’s say you give out a survey asking people to “select all that apply” from a short list. How do you visualize the results? If you make a simple bar chart showing the frequency of each option, you lose all of the relational data; you can no longer tell if , for example, most people who chose option 1 also chose option 3, but people who chose option 2 tend to only select that one. We set out to (partially) solve that problem with a graph (in the mathematical sense, as in “graph theory”). Nodes on the graph represent options and have an associated weight that corresponds to how many chose that option. Edges have a weight that represents the number of responses that chose both of the options it connects. The weight is visualized as a diameter or thickness.
We had part two of the C++ lesson next, which I hurried through so I could keep working on the DataViz code. I finished the part that uses numbers to create a picture, but it doesn’t find those numbers from the data. I need a second program that finds the prevalence of each option and pair so that I can use those numbers to make the final figure. Here’s an example with a 6-part “select all that apply” question using random numbers instead of real data, since I don’t have a way to sift through real data yet:
This lets us quickly see that everyone who selected the one represented in the lower right also selected the one in the upper left (since the widths of the circle and line are equal). We can also see that the one in the lower left was fairly popular, yet uncommon to see at the same time as the one in the upper right. I think my progress on programming this is going quite well and it’s nice to have some tangible progress on the overall mini project this week.
Anyway, it rained really hard when we went home today. Lots of lightning and wind and even some hail. I suspect my shoes will be drying for a few days. Thankfully, nothing got water damage. I heard a few complaints about pain from the hail, though.
Once home, I ate, blogged, and played a game (Jamestown) with all of my roommates.