18.8 Meeting Videos

18.8.1 Cohort 1

Meeting chat log
00:03:12    tan_iphone: Hullo!
00:05:15    tan_iphone: I seem to have influenced the ballot ok tho!
00:06:20    tan_iphone: Boaty mc boat dog
00:13:56    Jon Harmon (jonthegeek):    https://cran.r-project.org/package=nflfastR
00:17:17    Tony ElHabr:    Acronyms are my fav
00:17:21    Tony ElHabr:    *bacronyms
00:17:31    Jon Harmon (jonthegeek):    I think it stands for "I'm a Doctor Who fan."
00:17:59    Jordan Krogmann:    they have the death bots as the logo
00:18:10    Jordan Krogmann:    or whatever... I am a casual doctor fan
00:18:28    Jon Harmon (jonthegeek):    https://dalex.drwhy.ai/
00:24:26    Jon Harmon (jonthegeek):    Rowers in a boat is the explanation I've heard a lot. You can't just be the fastest, you need to also be in sync with the other rowers.
00:24:52    Tony ElHabr:    Is the default shapley? Or is it permutation? And what’s the difference?
00:31:19    Jon Harmon (jonthegeek):    Dunno about anybody else but I had to look up what that meant: ceteris paribus = "all other things being equal."
00:31:38    Tony ElHabr:    This is the same as partial dependence plots?
00:32:45    Tony ElHabr:    Oh, CP is the instance-level equivalent of PDP (used for data-set level)
00:33:46    Jon Harmon (jonthegeek):    "A profile showing how an individual observation’s prediction changes as a function of a given feature is called an ICE (individual conditional expectation) profile or a CP (ceteris paribus) profile."
Yeah, they never definitively explain PDP vs these other terms. I hate to make them rebuild this monster chapter but I think I have thoughts about making some stuff in here clearer!
00:37:52    Tony ElHabr:    i don’t think vip supports catboost 😢
00:39:04    Jon Harmon (jonthegeek):    Wait, do you pronounce it "vip" or "V I P"?
00:39:20    Tony ElHabr:    Yeah v i p
00:39:46    tan_iphone: very important predictor
00:40:04    tan_iphone: But wait is there a prequel?
00:40:56    Tony ElHabr:    meme (tweet) for the day: https://twitter.com/benbbaldwin/status/1431300791516663816?s=20
00:42:08    Jon Harmon (jonthegeek):    Dude, you can't send something that requires audio in the middle of a meeting 🙃
01:05:41    tan_iphone: Lmaoooo I have not nearly read enough of the book to do it
01:05:47    tan_iphone: Currently at the gym :) cheers folks

18.8.2 Cohort 3

18.8.3 Cohort 4

Meeting chat log
00:22:02    Federica Gazzelloni:    this is interesting: “ game theory techniques like SHAP are able to provide insight into how much crime is “good,” and how much is “too much.” (https://geophy.com/insights/whats-the-shap-how-crime-influences-property-value/)
00:26:51    Federica Gazzelloni:    example of using shap:
00:26:56    Federica Gazzelloni:    https://juliasilge.com/blog/board-games/