GioCirco

I'm not an AI-apologist, but 50% isn't out of the realm of possibility. I have worked on some tasks involving text-heavy processing that used to take 30+ minutes down to 5-6 minutes. It is highly task-specific.

This is the kind of hard-hitting science that I come to this site for.

The difference between models, drive-time vs fatality edition Easily one of the most common critiques I make when reviewing peer reviewed papers is the concept, the difference between statistically significant and not statistically significant is not itself s…

New blog post, in which I discuss the error of "the difference between stat significant and not is not itself stat significant". It often causes people to post-hoc try to explain things that are easily just due to the standard error of estimates, andrewpwheeler.com/2025/07/28/t...

Love the visual! I've tried to explain the advantages of splines over linear (or categorical) options a ton of times.

Is the categorical age different groups of ages binned into discrete categories? Regardless, a pretty good visualization of what information is lost in the transformation.

TRAJECTORIES OF CRIME AT PLACES: A LONGITUDINAL STUDY OF STREET SEGMENTS IN THE CITY OF SEATTLE* Studies of crime at micro places have generally relied on cross-sectional data and reported the distributions of crime statistics over short periods of time. In this paper we use official crime data ...

This is one of the "seminal" pieces in Crim about crime trajectories, and they find 18(!) different trajectories. Technically, they kept finding more, but cut the process off early:

onlinelibrary.wiley.com/doi/pdf/10.1...

These models have been very popular in Criminology historically, and I have similarly been pretty unimpressed with their performance.

This is pretty funny. I used to teach ggplot like it was making a pizza

AMA OLS vs Poisson regression Crazy busy with Crime De-Coder and day job, so this blog has gone by the wayside for a bit. I am doing more python training for crime analysts, most recently in Austin. If you want to get a flavor …

I recently received a question on Poisson vs OLS models for a dose response relationship, and posted the exchange to my blog. Long story short even with count data OLS models can make sense, it depends on the functional form.

andrewpwheeler.com/2025/05/28/a...

hit me up with that font

Also, if you are making cocktail syrups that have fat in them (like Orgeat) a bit helps keep everything in emulsion!

A plea for open access to qualitative criminology: With a Python script for anonymizing data and illustrative analysis of error rates This is the online version of the article. To access a print version with page numbers for citation and reference purposes, select

Paper with Scott Jacques on stripping sensitive information from narratives is out, www.qualitativecriminology.com/pub/zhiuy6jg.... Ultimate goal to make it easier for qual people to share there data for replication. Has links to python code and uses open source models.

Tags: "Fantasy"

The biggest findings were that - yes, we do see a relationship between alcohol-related offenses and gun crime BUT these are much smaller in effect compared to involvement in other crimes (e.g. motor vehicle theft, robbery, weapons possession).

Drunk and dangerous? Exploring the tenuous links among drunk driving, alcohol arrests, and firearm violence in an urban context Recent research and policy discussions have focused on prohibiting individuals with repeat alcohol-related offenses from purchasing or possessing fire…

After a long, long wait (and a transition from academia to private sector), our paper is finally out! www.sciencedirect.com/science/arti...

We used a unique source of data tracking residents of Milwaukee from the 1980s through the 2020s to examine the role of alcohol abuse on gun violence.

Got to love the speed of academic publishing. A paper that I helped collaborate on in 2021 is, just now, getting published. A full 4 years later, and also I am no longer in academia.

But tack on another +1 to the GoogleScholar page I guess!

You have displeased Miyazaki.

I never get tired of seeing this one.

Just ask 'lil Jon

The first time I ran across this ( on LinkedIn off all places) my first thought was "so you just invented regression adjustment?"

The secret of NIMH

And I have done both - to be clear. In a lot of cases there is diminishing utility of very advanced prediction software beyond using prior crime counts.

I would argue that beyond semantics there is really no clear difference. They might take a different logical approach, but like 99% of hot spot policing is just counting crimes at grid cells. And most of the "predictive policing" software relies on prior counts of crimes as well.

Looking forward to this game being good in 2-4 years!

I spend all day doing machine learning data pipelines and my body learns for stuff like this.

Ah yes, this drowning risk cluster is *checks notes* the entire nation of Iran.

Menu - Naing Myanmar Family Restaurant Lansing, MI 48910 Menu - Naing Myanmar Family Restaurant

I don't live in Lansing anymore, but the one place I miss is Naing Myanmar Family Restaurant: naing-myanmar-family-restaurant.res-menu.com/menu

I've been tons of places all over the world, and this is still some of the best food I've had. Truly great stuff.

For work outside of academia, Python is increasingly necessary. The library support and flexibility makes it useful in a lot of places. R does have better native support for some things (I like 'sf', 'ggplot' and some tidyverse stuff better for dataframes), but is not as good for a lot else.

Ahem, this is Apple *intelligence*

In general, I thought it was a nice approach to learn a bit more hands-on about poll aggregation, as well as remembering some of the distinct benefits of HLMs (e.g partial pooling, shrinkage).

A line graph showing the polling averages rating of Trump and Harris, with the red line representing Trump and the blue line representing Harris.

Remember the 2024 election? Are you NOT sick of hearing predictions about who will win? Want to revisit 11/4?!

Probably not. But I wrote up some thoughts on my novice approach to building a poll aggregation model during this last cycle. gmcirco.github.io/blog/posts/p...