Meeting chat log
00:14:27 Jonathan Trattner: Is there supposed to be sound?
00:16:42 Jonathan Trattner: I love how R’s parent S is up next
00:24:03 Asmae Toumi: Amazing
00:24:49 Scott Nestler: This whole chapter reminds me of a classic 2010 paper from my professor and friend Galit Shmueili, "To Explain or to Predict?" https://projecteuclid.org/journals/statistical-science/volume-25/issue-3/To-Explain-or-to-Predict/10.1214/10-STS330.full
00:25:48 Scott Nestler: Correction … her last name is Shmueli (had an extra 'i' in there).
00:26:08 Andy Farina: This chapter reminded me of a quote I have heard numerous times from my advisor over the past few years…”All models are wrong, some are useful”
00:26:24 Jon Harmon (jonthegeek): Hehe, yup!
00:28:11 Ben Gramza: I seem to hear that George Box quote 10000000 times a year
00:28:42 Conor Tompkins: map() go brrrrr
00:30:25 Conor Tompkins: Would be cool to create a raster of the density of the points, and find the differences between models
00:30:45 Jon Harmon (jonthegeek): I'm a little skeptical, we'll have to dig into that code!
00:31:30 Scott Nestler: Something doesn't seem right. Some of these metrics should *not* come up with the same results.
00:35:49 Conor Tompkins: Can you expect higher RMSE in general for higher priced homes? Ie $3 million. Is it better to use a percentage error term if that is the case?
00:54:24 Jonathan Trattner: Gotta head out now. Thanks Joe! Great job!
00:58:20 Asmae Toumi: Cool thanks
00:59:54 Tyler Grant Smith: like rock is correct pronunciation
01:02:10 Scott Nestler: A macro-average will compute the metric independently for each class and then take the average (hence treating all classes equally), whereas a micro-average will aggregate the contributions of all classes to compute the average metric.
01:02:21 Scott Nestler: The weighted macro computes them independently, but weights them by number of observations, rather than equally. Usually better than regular macro when there are class imbalances.
01:06:47 Andy Farina: Excellent Joe, thank you!