Andy Wheeler

Fitting a plateau effects model in scipy Dealing with a few models recently that people fit non-linear effects (either via polynomials or splines), and the results are just on their face too curvy. There is also a common social science tr…

Reminded of this old blog post in a thread on X, if you want to know how to fit a plateau in a regression model, this shows one way to constrain the coefficients and fit the model in scipy.

andrewpwheeler.com/2022/05/02/f...

Schmid also has a book on creating graphics by hand. Incredibly hard work and display of craftmanship.

Historical hot spot map of crimes in Seattle

Reminded of this old post, as Chris Bruce is doing a talk on historical overview of the crime analysis field.

Seattle hotspots are basically unchanged in 100 years (Schmid has a few in the 20's).

andrewpwheeler.com/2015/06/12/f...

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...

Asking ChatGPT to create an SVG knife from an example icon, and it messes up quite badly.

For an example much simpler than a pelican riding a bicycle, both Claude and ChatGPT cannot create an SVG knife.

Cost-benefit analysis of Gun Shot Detection Tech Is gun shot detection tech worth the cost?

Is gun shot detection tech worth the cost? I give an example cost-benefit analysis in this blog post, discussing lives saved. There is no consistent evidence at this point GSD reduces crime or increases clearances.

crimede-coder.com/blogposts/20...

Some notes on project management Have recently participated in several projects that I think went well at the day gig – these were big projects, multiple parties, and we came together and got to deployment on a shortened timeline.…

New blog post on project management, andrewpwheeler.com/2025/07/20/s...

High level goal make a list of things to-do and do them in a reasonable amount of time. Everything you do should keep these high level goals in mind.

Oh the problem is not limited to Crim (the clearly shady examples at GSU I found were in business school and one music prof). Those examples students cumulatively spend 100k per year in a class.

To be clear, these are examples profs self publishing, not just requiring an expensive textbook.

Ditto. Because these practices are not illegal, me and Scott are not sure where to go to next. Been trying to pitch to uni's to hire us to do a report. We may have to start writing congressmen (student grants do pay for some of them).

Most programs turn a blind eye.

How much do students pay for textbooks at GSU? Given I am a big proponent of open data, replicable scientific results, and open access publishing, I struck up a friendship with Scott Jacques at Georgia State University. One of the projects we p…

Me and Scott Jacques have done a similar analysis looking at book prices at GSU, andrewpwheeler.com/2025/02/06/h...

If you want analysis of this for your institution, feel free to get in touch.

Investigations: The School of Music and Dance’s half million dollar question - Daily Emerald A professor within the University of Oregon’s School of Music and Dance has earned roughly $570,000 in revenue from students in the past four years with his self-published book. However the issue was ...

Another example of a prof selling their own self published (likely mediocre) book to thousands of students. The article does not say online, but this prof may have made over 100k profit per year from sales.

dailyemerald.com/152165/featu...

Using LLMs to Extract Data from Text Using LLMs to Extract Data from Text

New blog post, I show how to use LLMs to extract pieces of information from plain text, with an example of extracting out info like modus operandi from police narratives

crimede-coder.com/blogposts/20...

Data Science for Crime Analysis with Python Data Science | Crime Analysis | python | book

Purchases for the Data Science for Crime Analysis with Python continue to trickle in, crimede-coder.com/blogposts/20...,

One of the reasons I promote python is because python jobs are much more common in private sector. It is IMO a better personal investment than R.

GSU Book Prices GSU Book Prices Tables

Weekend project, using R2 + DuckDB wasm to build a low price frontend for big data analytics crimede-coder.com/graphs/GSUBo...

Finding outliers in proportions Finding outliers in proportions

New blog post, finding outliers in proportions crimede-coder.com/blogposts/20...

Useful for flagging officers in early intervention systems, or examining data with varying denominators.

Blog_Code/Python/SequentialDec/SequentialDecision.ipynb at master · apwheele/Blog_Code Code associated with my blog posts. Contribute to apwheele/Blog_Code development by creating an account on GitHub.

Nerdsniped by a sequential decision problem on X, wrote up a few different solutions + simulations in python github.com/apwheele/Blo...

Working on databases from prison: How I got here, part 2.

Interesting post of self teaching coding and now working in software dev in prison, turso.tech/blog/working...

Build Stuff I have had this thought in my head for awhile – criminology research to me is almost all boring. Most of the recent advancement in academia is focused on making science more rigorous – more open me…

New blog post, academic criminologists should be more focused on building stuff than writing papers, andrewpwheeler.com/2025/06/12/b...

Denver's STAR Program and Disorder Crime Reductions Denver's STAR Program and Disorder Crime Reductions

With the new ASEBP site, my Criminal Justician essays were not transferred over. Putting them up on my crime de-coder site. First one is description why I am skeptical of a study of Denver alt response having large crime reductions, crimede-coder.com/blogposts/20....

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...

One of the projects on the forever bucket list is to try to use police data (victimizations and offenses of homeless individuals) and see how well that spatially correlates with PIT or your tent measure here.

Besides the fact you could almost do this yourself anymore (you can post pre-prints with self-cites and they get picked up). My personal blog is picked up by Google Scholar.

Blog cited around ~10 times in real papers, and it had references in the posts. Then GS crawled rest of my site.

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.

Please make your papers open access: there really are people out there who want to read them but don’t have access to subscription journals.

Fortunately, making your work open access is both very easy and 100% free: www.lesscrime.info/post/how-to-...

YouTube video by CrimeDe-Coder Tutorial on Using Quarto + VS Code

Most of my personal work on creating reports I have migrated from Jupyter notebooks to Quarto. Here is a video tutorial I made for Kansas City crime analysts python training using Quarto + VS Code.

www.youtube.com/watch?v=gaPP...

Are you interested in designing network experiments where you control the number of direct treated vs spillover? Check out these functions I have written in python to optimally spread out treatment in your network given constraints

github.com/apwheele/Blo...

How to get property fraud alerts using Wake County’s new land-use records system Deed fraud is rare in Wake County. However, it’s come under scrutiny after a recent high-profile case in Raleigh.

It has happened enough times in triangle area in NC for the local paper to have several stories, www.newsobserver.com/news/busines...

www.newsobserver.com/news/local/c...

Robberies in Houston by month going down

Burglaries in Phoenix by month decreasing

One of the ideas I had, but have never put the details to paper, is a monitoring system for crimes that looks at the relative mix. Mostly for theft related crimes, to see if displacement occurs from one type to another.

There are all sorts of large scale changes in US.

Crime Analysis and Mapping: Professional Training in ArcGIS Pro · CrimRxiv

Eric Piza has a nice set of courses on crime mapping using ArcGIS in his grad program at Northeastern

www.crimrxiv.com/northeastern...

Using Esri + python: arcpy notes I shared a series of posts this week using Esri + arcpy tools on my Crime De-Coder LinkedIn page. LinkedIn eventually removes the posts though, so I am putting those same tips here on the blog. Esr…

I put my notes on using python + ArcGIS (mostly have been posting on LinkedIn) on my blog.

Mostly for the bots at this point to read and have better examples.

andrewpwheeler.com/2025/04/20/u...