Mapping My Life; Or, Wow I Need to Get Out of Oakland
Our group decided to bend the rules a bit and create a data autobiography of our whole group. We wanted to map our locations to see where we went in a week, what locations were frequented by which people, and whether, when put together, our location data would tell a meaningful story. We downloaded a fleet of GPS tracking apps and figured out which ones worked best to record an accurate location every ten minutes, in a format we could export to csv. We recorded for about a week, then worked to represent the data on a map of Pittsburgh and draw conclusions from this visualization.
My part in the group took place mostly during data collection and manipulation. I helped to collect the gpx files that each phone collected, used an xml editor to make them agree with each other, and created a csv of our group’s data. Many of the problems of our data came out of the differences between function when using the Android and Apple versions of the GPS Logger app that we chose. It led to a large difference in the number of points collected. On Android, it was possible to choose how frequently the location would be taken, which is why Connor’s data is so large: he was initially recording his location every minute. For these purposes, Maya’s data is the most consistent. On Apple, the recording was much less frequent. It seemed to me like it only took your location when the location significantly changed. When I noticed this, I started collecting additional data from another app we considered, that recorded the location every ten minutes but had to be exported about every two hours or it would delete itself. Much of my time was spent combining these two sources, and I wish I had used a dataframe or regular expressions to work with that, instead of diving right into the spreadsheets and gpx files, as it would have saved me a lot of work. My personal data ended up sort of bridging the difference between the Apple and Android set, but if we were doing any quantitative analysis instead of qualitative or visual, I would need to fix these differences, either by deleting many of the duplicate points recorded by the Android phones, or using percentages instead. Alternatively, with more time to test different sources and perhaps paid apps instead of free, we could have a more accurate and consistent dataset.
After compiling gpx and cvs files of the whole group, I started to plug them into google maps just to have fun and see what our data would look like. This also told an interesting story about the inaccuracies of our locations. Jahari’s data may look plausible when presented as a collection of points, but as a track that represents the data chronologically, he makes many impossible leaps from place to place, leaving his visualization looking spiky while most of the other tracks move in circular routes. We’re not sure what went wrong, but his app pinged him in several places he’d never been. While Jahari is the most drastic case, it happened with several of our own datasets, across both Apple and Android apps. Also, when you think about the context that gps devices are used in, the errors take on new meaning. If a helicopter parent was tracking their child, someone might get unfairly grounded. One would hope that in more official settings, the gps tool used would be more accurate to avoid more troubling consequences.
In the rest of the process I largely took a backseat because my group members are so much more advanced at code. I paid attention and largely attempted to understand what they were doing, but other than that I mostly provided moral support and occasional ideas and votes on format and design.
Now that our data is compiled and successfully visualized, certain implications of our data collection began to surface. The map forces you to think about patterns in your life. My week was such a very typical week for me that I find myself thinking I need to branch out of my routine once or twice in my life. On a slightly more alarming level, the places I live, frequent, and travel, when illustrated as a set of points and as a route, are very obvious. Even with just old data, it would be easy to guess where I would be at any time, so it’s rather lucky it's the end of the year and I won't be walking these trails anymore. But overall I was very happy to be a part of this project due to how well it came together and how beautifully our data visualization and site has turned out.