September 21st weekend

I’ve been watching this YouTuber recently called Quantum Boy. His videos are poetry to me, both in his video editing style and in how much I relate to him personally. He’s currently a Physics PhD student in Boston, and seeing him explore the city reminded me of the nostalgic and wonder-filled lens through which I viewed the world around me during my early days in Pittsburgh. Maybe it’s the subway heat and grime, but that’s not the way I view New York. It could also be that this lens gets applied with both time and distance, but that’s not something I can tell for a while later. I definitely don’t regret any way that I’ve used my time here so far, but if there is one thing I do regret, it is not developing that sense of attachment to the city. That being said, I may or may not be leaving for a different city in a year, so this is something that I’ll be keeping in mind going into grad school. Outside of that, though, in watching Quantum Boy, I’ve realized that no matter what, the one thing I do want in my new place next year is a blackboard installed into the wall (and some Hagoromo chalk to go with it too).

In other news, I went to see a Miami FC vs. New York FC game today. To date, this has been the sporting event with the most energy that I’ve been to, which seems crazy to me given that it is only MLS (Major League Soccer). I’ve always passively had some desire to go to a premier league football game but this experience has pushed that desire to the forefront. I can’t imagine what it would feel like at some Dortmund game (which happens to be the first jersey I’ve ever gotten) with a massive tifo in the back.

I’m going to close off by addressing the lack of econ in my posts recently. I actually have a number of drafts in the making but finding the time to write each post to my personal satisfaction is quite difficult right now amidst grad apps and NSF. I will say, though, that there are quite a few bangers in store so be on the lookout for those for whenever I get the time to be able to flesh them out. The one enormous victory I’ve had in terms of finishing a project, however, is TidyTuesday. I know I said that I would have it up a lot earlier but I hadn’t finished the latex automated report building by then and the thought of having to latex everything in a nicely formatted manner only to have to redo it while writing python code to write in latex disgusted me so I put it off for a while.

I’ve managed to finally finish the latex automation last Wednesday so I’ve been cooking up these random regressions for the past handful of days to prep a post this coming Tuesday. Since I have to do minimal work for TidyTuesday going forward, get ready to finally see posts every Tuesday (even while apps are going on). I’ve talked about what my TidyTuesday project is a few times before but I feel the need to explain again in each new post about it. Basically it is a repo that takes two random datasets every day from some API, runs a linear regression, and then creates a plot, regression table, and writes some paragraphs explaining each of the time series used in the regression. So far I only pull data using the FRED API (and yes I mean FRED (Federal Reserve Economic Data) instead of FED, I’ve had some questions about this) but I’ve been working with the Bureau of Economic Analysis (BEA) and Bureau of Labor Statistics (BLS) APIs as well recently so that’s a next step. One other next step is making my regressions smarter in a way that doesn’t require me to actively do work. In other words, running linear regressions between random datasets doesn’t usually really yield any beneficial insight but often times this is also because the relationship between any given two series can be non-linear or there are some obvious confounding variables that delegitimizes any insight we could glean. For Time series especially, one often times detrends when creating models which is not something that I am currently doing. Building out this aspect of the code is a bit more nuanced of a problem; I have a few ideas in mind such as keeping a list of all the various regression techniques and choosing whichever one fares the best (by some metric that I decide) or even asking some chat gpt like AI for its recommendation on what type of regression to run and just running that. Regardless, this is a problem for future Pranay who is already done with his applications but I’m curious about the potential of this idea.

I don’t have any underlying purpose driving this TidyTuesday project other than it seemed like something interesting to do at the time and recently I’ve been feeling like that is something missing from academia (at least in econ). Not to say that pursuing research problems with a strong motivation or noble purpose is wrong, I just feel like doing something just for the sake of it being an interesting problem to consider should be more acceptable.