back
user profile
apwheele
1,887karma·374submissions·November 10, 2021
about
Data scientist. Former academic in criminal justice field.
Personal blog at https://andrewpwheeler.com/
Consulting at https://crimede-coder.com/
Large Language Models for Mortals: A Practical Guide for Analysts (book), https://crimede-coder.com/blogposts/2026/LLMsForMortals
recent activity (374 total)
comment
Reminds me a bit of this blog post by Tal Yarkoni, https://www.talyarkoni.org/blog/2018/08/18/if-we-already-und... . That is perhaps more consistent with the "…
comment
I found it resonated (although agree the title is click-baity, they just talk about their own experiences in multiple orgs). It is hard to do good data work. Offhand it takes some combination of: - bu…
comment
For an alt take, https://grahamfactor.substack.com/p/qualified-immunity-in-co... …
comment
Due to the curse of dimensionality, more of the points are pushed toward the hull in higher dimensions. So I don't think this is likely to be very effective as a data exploration technique as sta…
comment
Just as an alt perspective, gained quite a bit of weight in mid 20's-30's. I tried various things -- I did Judo in college before gained weight, but doing BJJ after I gained weight I could n…
comment
Agree with this, the data is consistent with small, additive, effects of learning over time. (And those additive effects don't show much variance between students, at least for these short course…
comment
Pretty much. Some things to ease the process are to make the questions asked specific as possible (no good if the question can be ambiguous sort of in the end), and to have feedback (plus initial trai…
comment
At work, we have a team that is working on various agile dashboards for project metrics (acceleration or burndown or whatever, all the stuff I would guess the majority of folks commenting on Hacker Ne…
comment
One I have seen for time-series forecasting is the inverse-hyperbolic sine function as well, https://robjhyndman.com/hyndsight/transformations/ .…
comment
For a random python/pandas trick, I have come across web-api's that cannot be directly read into pandas using the URL (I imagine folks on here can comment better the difference in web servin…
comment
I will have to give these formulations a try. IMO pretty common to have asymptote like theoretical behavior in the tails with social science data, https://andrewpwheeler.com/2022/…
comment
While it is true that getting a traditional tenure track job is very tough in the current market, programs like the one in the article can make modest changes to better prep students for other roles. …
comment
Yeah, and the R ecosystem has this built in (checking the examples run, not their output is correct). `R CMD check` has as one of its checks whether the examples you build in the help docs generate ru…
comment
The way that these programs work, it is no questions asked (intended to get even criminals to turn in firearms, to prevent them from using in future crimes/suicides). So a "show ID to turn i…
comment
That was part of the joke (I meant macro on the right hand side, not micro), but will concede I am not a terribly funny person. I think micro is much easier/more intuitive personally (I think you…
comment
I think this is plausible on the margins, although it could be true that Davidson relative to peers are lower but still overall take the same test they are higher compared to 20 years ago as well (is …
comment
Utility(Sleeping In) > Utility(A in MacroEconomics)
comment
Also recommend Hubbard's How to Measure Anything . Background in social science & stats, so was familiar with the decision making logic before reading, but think the book is more straight fo…
comment
Nice job -- I have examples using statsmodels for similar (not time series) data [1,2]. I typically use this for EDA before regression modelling, so dependencies in that scenario are not a big deal. B…
comment
FYI for folks if you want to convert jupyter notebooks to html you can use nbconvert: jupyter nbconvert --execute example_report.ipynb --no-input --to html
I have never been able to get a nic…
comment
End users even have problems interpreting probability statements in my experience (e.g. this factor increases probability of event by 10%) let alone graphs. When you get more complicated models, those…
comment
Yeah, currently I do things like "Top 5 important factors" for shapely values, but have debated on even doing that much. There is good evidence these reduced form estimates can often be misl…
comment
Ditto, although for my applications they are not live (typical business processes lag by a month or more). So most of my stakeholders are happy that I just send a monthly PDF with updates. The endless…
comment
I picked up this book and enjoyed it just for the process, https://www.amazon.com/Make-Your-Own-Pixel-Art/dp/1593278861 . (Not an artist, just a programmer who has tried to m…