All posts in category data science

Using DuckDB WASM + Cloudflare R2 to host and query big data (for almost free)

The motivation here, prompted by a recent question Abigail Haddad had on LinkedIn:

For the machines, the context is hosting a dataset of 150 million rows (in another post Abigail stated it was around 72 gigs). And you want the public to be able to make ad-hoc queries on that data. Examples where you may want to do this are public dashboards (think a cities open data site, just puts all the data on R2 and has a front end).

This is the point where traditional SQL databases for websites probably don’t make sense. Databases like Supabase Postgres or MySQL can have that much data, given the cost of cloud computing though and what they are typically used for, it does not make much sense to put 72 gigs and use them for data analysis type queries.

Hosting the data as static files though in an online bucket, like Cloudflare’s R2, and then querying the data makes more sense for that size. Here to query the data, I also use a WASM deployed DuckDB. What this means is I don’t really have to worry about a server at all – it should scale to however many people want to use the service (I am just serving up HTML). The client’s machine handles creating the query and displaying the resulting data via javascript, and Cloudflare basically just pushes data around.

If you want to see it in action, you can check out the github repo, or see the demo deployed on github pages to illustrate generating queries. To check out a query on my Cloudflare R2 bucket, you can run SELECT * FROM 'https://data-crimedecoder.com/books.parquet' LIMIT 10;:

Cloudflare is nice here, since there are no egress charges (big data you need to worry about that). You do get charged for different read/write operations, but the free tiers seem quite generous (I do not know quite how to map these queries to Class B operations in Cloudflare’s parlance, but you get 10 million per month and all my tests only generated a few thousand).

For some notes on this set-up. On Cloudflare, to be able to use DuckDB WASM, I needed to expose the R2 bucket via a custom domain. Using the development url did not work (same issue as here). I also set my CORS Policy to:

[
  {
    "AllowedOrigins": [
      "*"
    ],
    "AllowedMethods": [
      "GET",
      "HEAD"
    ],
    "AllowedHeaders": [
      "*"
    ],
    "ExposeHeaders": [],
    "MaxAgeSeconds": 3000
  }
]

While my Crime De-Coder site is PHP, all the good stuff happens client-side. So you can see some example demo’s of the GSU book prices data.

One of the annoying things about this though, with S3 you can partition the files and query multiple partitions at once. Here something like SELECT * FROM read_parquet('https://data-crimedecoder.com/parquet/Semester=*/*') LIMIT 10; does not work. You can union the partitions together manually. So not sure if there is a way to set up R2 to work the same way as the S3 example (set up a FTP server? let me know in the comments!).

For pricing, for the scenario Abigail had of 72 gigs of data, we then have:

  • $10 per year for the domain
  • 0.015*72*12 = $13 for storage of the 72 gigs

So we have a total cost to run this of $23 per year. And it can scale to a crazy number of users and very large datasets out of the box. (My usecase here is just $10 for the domain, you get 10 gigs for free.)

Since this can be deployed on a static site, there are free options (like github pages). So the page with the SQL query part is essentially free. (I am not sure if there is a way to double dip on the R2 custom domain, such as just putting the HTML in the bucket. Yes, you can just put the HTML in the bucket and it will render like normal.)

While this example only shows generating a table, you can do whatever additional graphics client side. So could make a normal looking dashboard with dropdowns, and those just execute various queries and fill in the graphs/tables.

5 Comments
by Andy Wheeler on June 29, 2025  •  Permalink
Posted in data science, website
Tagged ,

Posted by Andy Wheeler on June 29, 2025

https://andrewpwheeler.com/2025/06/29/using-duckdb-wasm-cloudflare-r2-to-host-and-query-big-data-for-almost-free/

Build Stuff

I have had this thought in my head for a while – criminology research to me is almost all boring. Most of the recent advancement in academia is focused on making science more rigorous – more open methods, more experiments, stronger quasi-experimental designs. These are all good things, but to me still do not fundamentally change the practical implementation of our work.

Criminology research is myopically focused on learning something – I think this should be flipped, and the emphasis be on doing something. We should be building things to improve the crime and justice system.

How criminology research typically goes

Here is a screenshot of the recent articles published in the Journal of Quantitative Criminology. I think this is a pretty good cross-section of high-quality, well-respected research in criminology.

Three of the four articles are clearly ex-ante evaluations of different (pretty normal) policies/behavior by police and their subsequent downstream effects on crime and safety. They are all good papers, and knowing how effective a particular policy works (like stop and frisk, or firearm seizures) are good! But they are the literal example where the term ivory tower comes from – these are things happening in the world, and academics passively observe and say how well they are working. None of the academics in those papers were directly involved in any boots on the ground application – they were things normal operations the police agencies in question were doing on their own.

Imagine someone said “I want to improve the criminal justice system”, and then “to accomplish this, I am going to passively observe what other people do, and tell them if it is effective or not”. This is almost 100% of what academics in criminology do.

The article on illicit supply chains is another one that bothers me – it is sneaky in the respect that many academics would say “ooh that is interesting and should be helpful” given its novelty. I challenge anyone to give a concrete example of how the findings in the article can be directly useful in any law enforcement context. Not hypothetical, “can be useful in targeting someone for investigation”, like literal “this specific group can do specific X to accomplish specific Y”. We have plenty of real problems with illicit supply chains – drug smuggling in and out of the US (recommend the contraband show on Amazon, who knew many manufactures smuggle weed from US out to the UK!). Fentanyl or methamphetamine production from base materials. Retail theft groups and selling online. Plenty of real problems.

Criminology articles tend to be littered with absurdly vague accusations that they can help operations. They almost always cannot.

So we have articles that are passive evaluations of policies other people thought up. I agree this is good, but who exactly comes up with the new stuff to try out? We just have to wait around and hope other people have good ideas and take the time to try them out. And then we have theoretical articles larping as useful in practice (since other academics are the ones reviewing the papers, and no one says “erm, that is nice but makes no sense for practical day to day usage”).

Some may say this is the way science is supposed to work. My response to that is I don’t know dude, go and look at what folks are doing in the engineering or computer science or biology department. They seem to manage both theoretical and practical advancements at the same time just fine and dandy.

Well what have you built Andy?

It is a fair critique if you say “most of your work is boring Andy”. Most of my work is the same “see how a policy works from the ivory tower”, but a few are more “build stuff”. Examples of those include:

In the above examples, the one that I know has gotten the most traction are simple rules to identify crime spikes. I know because I have spent time demonstrating that work to various crime analysts across the country, and so many have told me “I use your Poisson Z-score Andy”. (A few have used the patrol area work as well, so I should be in the negative for carbon generation.)

Papers are not what matter though – papers are a distraction. The applications are what matter. The biggest waste currently in academic criminology work is peer reviewed papers. Our priorities as academics are totally backwards. We are evaluated on whether we get a paper published, we should be evaluated on whether we make the world a better place. Papers by themselves do not make the world a better place.

Instead of writing about things other people are doing and whether they work, we should spend more of our time trying to create things that improve the criminal justice system.

Some traditional academics may not agree with this – science is about formulating and testing hypotheses. This need not be in conflict with doing stuff. Have a theory about human nature, what better way to prove the theory than building something to attempt to change things for the better according to your theory. If it works in real life to accomplish things people care about guess what – other people will want to do it. You may even be able to sell it.

Examples of innovations I am excited about

Part of what prompted this was I was talking to a friend, and basically none of the things we were excited about have come from academic criminologists. I think a good exemplar of what I mean here is Anthony Tassone, the head of Truleo. To be clear, this is not a dig but a compliment, following some of Anthony’s posts on social media (LinkedIn, X), he is not a Rhodes Scholar. He is just some dude, building stuff for criminal justice agencies mostly using the recent advancements in LLMs.

For a few other examples of products I am excited about how they can improve criminal justice (I have no affiliations with these beyond I talk to people). Polis for evaluating body worn camera feeds. Dan Tatenko for CaseX is building an automated online crime reporting system that is much simpler to use. The folks at Carbyne (for 911 calls) are also doing some cool stuff. Matt White at Multitude Insights is building a SaaS app to better distribute BOLOs.

The folks at Polis (Brian Lande and Jon Wender) are the only two people in this list that have anything remotely to do with academic criminology. They each have PhDs (Brian in sociology and Jon in criminology). Although they were not tenure track professors, they are former/current police officers with PhDs. Dan at CaseX was a detective not that long ago. The folks at Carbyne I believe are have tech backgrounds. Matt has a military background, but pursued his start up after doing an MBA.

The reason I bring up Anthony Tassone is because when we as criminologists say we are going to passively evaluate what other people are doing, we are saying “we will just let tech people like Anthony make decisions on what real practitioners of criminal justice pursue”. Again not a dig on Anthony – it is a good thing for people to build cool stuff and see if there is a market. My point is that if Anthony can do it, why not academic criminologists?

Rick Smith at Axon is another example. While Axon really got its dominate market due to conducted energy devices and then body worn cameras (so hardware), quite a bit of the current innovation at Axon is software. And Rick did not have a background in hardware engineering either, he just had an idea and built it.

Transferring over into professional software engineering since 2020, let me tell my fellow academics, you too can write software. It is more about having a good idea that actually impacts practice.

Where to next?

Since the day gig (working on fraud-waste-abuse in Medicaid claims) pays the bills, most of my build stuff is now focused on that. The technical skills to learn software engineering are currently not effectively taught in Criminal Justice PhD programs, but they could be. Writing a dissertation is way harder than learning to code.

While my python book has a major focus on data analysis, it is really the same skills to jump to more general software engineering. (I specifically wrote the book to cover more software engineering topics, like writing functions and managing environments, as most of the other python data science books lack that material.)

Skills gap is only part of the issue though. The second is supporting work that pursues building stuff. It is really just norms in the current academe that stop this from occurring now. People value papers, NIJ (at least used to) mostly fund very boring incremental work.

I discussed start ups (people dreaming and building their own stuff) and other larger established orgs (like Axon). Academics are in a prime position to pursue their own start ups, and most Universities have some support for this (see Joel Caplan and Simsi for an example of that path). Especially for software applications, there are few barriers. It is more about time and effort spent pursuing that.

I think the more interesting path is to get more academic criminologists working directly with software companies. I will drop a specific example since I am pretty sure he will not be offended, everyone would be better off if Ian Adams worked directly for one of these companies (the companies, Ian’s take home pay, long term advancement in policing operations). Ian writes good papers – it would be better if Ian worked directly with the companies to make their tools better from the get go.

My friend I was discussing this with gave the example of Bell Labs. Software orgs could easily have professors take part time gigs with them directly, or just go work with them on sabbaticals. Axon should support something like that now.

While this post has been focused on software development, I think it could look similar for collaborating with criminal justice agencies directly. The economics will need to be slightly different (they do not have quite as much expendable capital to support academics, the ROI for private sector I think should be easily positive in the long run). But that I think that would probably be much more effective than the current grant based approach. (Just pay a professor directly to do stuff, instead of asking NIJ to indirectly support evaluation of something the police department has decided to already put into operation.)

Scientific revolutions are not happening in journal articles. They are happening by people building stuff and accomplishing things in the real world with those innovations.

For a few responses to this post, Alex sent me this (saying my characterization of Larry as passively observing is not quite accurate), which is totally reasonable:

Nice post on building/ doing things and thanks for highlighting the paper with Larry. One error however, Larry was directly involved in the doing. He was the chief science officer for the London Met police and has designed their new stop and frisk policy (and targeting areas) based directly on our work. Our work was also highlighted by the Times London as effective crime policy and also by the Chief of the London Met Police as well who said it was one of the best policy relevant papers he’s ever seen. All police are now being by trained on the new legislation on stop and search in procedurally just ways. You may not have known this background but it’s directly relevant to your post.

Larry Sherman (and David Weisburd), and their work on hot spots + direct experiments with police are really exemplars of “doing” vs “learning”. (David Kennedy and his work on focused deterrence is another good example.) In the mid 90s when Larry or David did experiments, they likely were directly involved in a way that I am suggesting – the departments are going and asking Larry “what should we do”.

My personal experience, trying to apply many of the lessons of David’s and Larry’s work (which was started around 30 years ago at this point), is not quite like that. It is more of police departments have already committed to doing something (like hotspots), and want help implementing the project, and maybe some grant helps fund the research. Which is hard and important work, but honestly just looks like effective project management (and departments should just invest in researchers/project managers directly, the external funding model does not make sense long term). For a more on point example of what I mean by doing, see what Rob Guerette did as an embedded criminologist with Miami PD.

Part of the reason I wrote the post, if you think about the progression of policing, we have phases – August Vollmer for professionalization in the early 1900’s. I think you could say folks like Larry and David (and Bill Bratton) brought about a new age of metrics to PDs in the 90s.

There are also technology changes that fundamentally impact PDs. Cars + 911 is one. The most recent one is a new type of oversight via body worn cameras. Folks who are leading this wave of professionalization changes are tech folks (like Rick Smith and Anthony Tassone). I think it is a mistake to just sit on the sidelines and see what these folks come up with – I want academic criminologists to be directly involved in the nitty gritty of the implementations of these systems and making them better.

A second response to this is that building stuff is hard, which I agree and did not mean to imply it was as easy as writing papers (it is not). Here is Anthony Tassone’s response on X:

I know this is hard. This is part of why I mentioned the Bell labs path. Working directly for an already established company is much easier/safer than doing your own startup. Bootstrapping a startup is additionally much different than doing VC go big or go home – which academics on sabbaticals and as a side hustle are potentially in a good position to do this.

Laura Huey did this path, and does not have nice things to say about it:

I have not talked to Laura specifically about this, but I suspect it is her experience running the Canadian Society of Evidence Based Policing. Which I would not suggest starting a non-profit either honestly. Even if you start a for-profit, there is no guarantee you will be in a good position in your current academic position to be well supported.

Again no doubt building useful stuff is harder than writing papers. For a counter to these though, doing my bootstrapped consulting firm is definitely not as stressful as building a large company like Anthony. And working for a tech company directly was a good career move for me (although now I spend most of my day building stuff to limit fraud-waste-abuse in Medicaid claims, not improving policing).

My suggestion that the field should be more focused on building stuff was not because it was easier, it was because if you don’t there is a good chance you are mostly irrelevant.

1 Comment
by Andy Wheeler on June 12, 2025  •  Permalink
Posted in Crime Analysis, Criminal Justice, data science, Python, scholarly
Tagged

Posted by Andy Wheeler on June 12, 2025

https://andrewpwheeler.com/2025/06/12/build-stuff/

LinkedIn is the best social media site

The end goals I want for a social media site are:

  • promote my work
  • see other peoples work

Social media for other people may have other uses. I do comment and have minor interactions on the social media sites, but I do not use them primarily for that. So my context is more business oriented (I do not have Facebook, and have not considered it). I participate some on Reddit as well, but that is pretty sparingly.

LinkedIn is the best for both relative to X and BlueSky currently. So I encourage folks with my same interests to migrate to LinkedIn.

LinkedIn

So I started Crime De-Coder around 2 years ago. I first created a website, and then second started a LinkedIn page.

When I first created the business page, I invited most of my criminal justice contacts to follow the page. I had maybe 500 followers just based on that first wave of invites. At first I posted once or twice a week, and it was very steady growth, and grew to over 1500 followers in maybe just a month or two.

Now, LinkedIn has a reputation for more spammy lifecoach self promotion (for lack of a better description). I intentionally try to post somewhat technical material, but keep it brief and understandable. It is mostly things I am working on that I think will be of interest to crime analysts or the general academic community. Here is one of my recent posts on structured outputs:

Current follower count on LinkedIn for my business page (which in retrospect may have been a mistake, I think they promote business pages less than personal pages), is 3230, and I have fairly consistent growth of a few new followers per day.

I first started posting once a week, and with additional growth expanded to once every other day and at one point once a day. I have cut back recently (mostly just due to time). I did get more engagement, around 1000+ views per day when I was posting every day.

Probably the most important part though of advertising Crime De-Coder is the types of views I am getting. My followers are not just academic colleagues I was previously friends with, it is a decent outside my first degree network of police officers and other non-profit related folks. I have landed several contracts where I know those individuals reached out to me based on my LinkedIn posting. It could be higher, as my personal Crime De-Coder website ranks very poorly on Bing search, but my LinkedIn posts come up fairly high.

When I was first on Twitter I did have a few academic collaborations that I am not sure would have happened without it (a paper with Manne Gerell, and a paper with Gio Circo, although I had met Gio in real life before that). I do not remember getting any actual consulting work though.

I mentioned it is not only better for me for advertising my work, but also consuming other material. I did a quick experiment, just opened the home page and scrolled the first 3 non-advertisement posts on LinkedIn, X, and BlueSky. For LinkedIn

This is likely a person I do not want anything to do with, but their comment I agree with. Whenever I use Service Now at my day job I want to rage quit (just send a Teams chat or email and be done with it, LLMs can do smarter routing anymore). The next two are people are I am directly connected with. Some snark by Nick Selby (which I can understand the sentiment, albeit disagree with, I will not bother to comment though). And something posted by Mindy Duong I likely would be interested in:

Then another advert, and then a post by Chief Patterson of Raleigh, whom I am not directly connected with, but was liked by Tamara Herold and Jamie Vaske (whom I am connected with).

So annoying for the adverts, but the suggested (which the feeds are weird now, they are not chronological) are not bad. I would prefer if LinkedIn had a “general” and “my friends” sections, but overall I am happier with the content I see on LinkedIn than I am the other sites.

X & BlueSky

I first created a personal then Twitter account in 2018. Nadine Connell suggested it, and it was nice then. When I first joined I think it was Cory Haberman tweeted and said to follow my work, and I had a few hundred followers that first day. Then over the next two years, just posting blog posts and papers for the most part, I grew to over 1500 followers IIRC. I also consumed quite a bit of content from criminal justice colleagues. It was much more academic focused, but it was a very good source of recent research, CJ relevant news and content.

I then eventually deleted the Twitter account, due to a colleague being upset I liked a tweet. To be clear, the colleague was upset but it wasn’t a very big deal, I just did not want to deal with it.

I started a Crime De-Coder X account last year. I made an account to watch the Trump interview, and just decided to roll with it. I tried really hard to make X work – I posted daily, the same stuff I had been sharing on LinkedIn, just shorter form. After 4 months, I have 139 followers (again, when I joined Twitter in 2018 I had more than that on day 1). And some of those followers are porn accounts or bots. Majority of my posts get <=1 like and 0 reposts. It just hasn’t resulted in getting my work out there the same way in 2018 or on LinkedIn now.

So in terms of sharing work, the more recent X has been a bust. In terms of viewing other work, my X feed is dominated by short form video content (a mimic of TikTok) I don’t really care about. This is after extensively blocking/muting/saying I don’t like a lot of content. I promise I tried really hard to make X work.

So when I open up the Twitter home feed, it is two videos by Musk:

Then a thread by Per-Olof (whom I follow), and then another short video Death App joke:

So I thought this was satire, but clicking that fellows posts I think he may actually be involved in promoting that app. I don’t know, but I don’t want any part of it.

BlueSky I have not been on as long, but given how easy it was to get started on Twitter and X, I am not going to worry about posting so much. I have 43 followers, and posts similar to X have basically been zero interaction for the most part. The content feed is different than X, but is still not something I care that much about.

We have Jeff Asher and his football takes:

I am connected with Jeff on LinkedIn, in which he only posts his technical material. So if you want to hear Jeff’s takes on football and UT-Austin stuff then go ahead and follow him on BlueSky. Then we have a promotional post by a psychologist (this person I likely would be interested in following his work, this particular post though is not very interesting). And a not funny Onion like post?

Then Gavin Hales, whom I follow, and typically shares good content. And another post I leave with no comment.

My BlueSky feed is mostly dominated by folks in the UK currently. It could be good, but it currently just does not have the uptake to make it worth it like I had with Twitter in 2018. It may be the case given my different goals, to advertise my consulting business, Twitter in 2018 would not be good either though.

So for folks who subscribe to this blog, I highly suggest to give LinkedIn a try for your social media consumption and sharing.

Leave a comment
by Andy Wheeler on February 20, 2025  •  Permalink
Posted in data science, Personal Productivity, scholarly, social networking, writing
Tagged ,

Posted by Andy Wheeler on February 20, 2025

https://andrewpwheeler.com/2025/02/20/linkedin-is-the-best-social-media-site/

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 pursued was a pretty simple, but could potentially save students a ton of money. If you have checked out your universities online library system recently, you may have noticed they have digital books (mostly from academic presses) that you can just read. No limits like the local library, they are just available to all students.

So the idea Scott had was identify books students are paying for, and then see if the library can negotiate with the publisher to have it for all students. This shifts the cost from the student to the university, but the licensing fees for the books are not that large (think less than $1000). This can save money especially if it is a class with many students, so say a $30 book with 100 students, that is $3000 students are ponying up in toto.

To do this we would need course enrollments and the books they are having students buy. Of course, this is data that does exist, but I knew going in that it was just not going to happen that someone just nicely gave us a spreadsheet of data. So I set about to scrape the data, you can see that work on Github if you care too.

The github repo in the data folder has fall 2024 and spring 2025 Excel spreadsheets if you want to see the data. I also have a filterable dashboard on my crime de-coder site.

You can filter for specific colleges, look up individual books, etc. (This is a preliminary dashboard that has a few kinks, if you get too sick of the filtering acting wonky I would suggest just downloading the Excel spreadsheets.)

One of the aspects though of doing this analysis, the types of academic publishers me and Scott set out to identify are pretty small fish. The largest happen to be Academic textbook publishers (like Pearson and McGraw Hill). The biggest, coming in at over $300,000 students spend on in a year is a Pearson text on Algebra.

You may wonder why so many students are buying an algebra book. It is assigned across the Pre-calculus courses. GSU is a predominantly low income serving institution, with the majority of students on Pell grants. Those students at least will get their textbooks reimbursed via the Pell grants (at least before the grant money runs out).

Being a former professor, these course bundles in my area (criminal justice) were comically poor quality. I accede the math ones could be higher quality, I have not purchased this one specifically, but this offers two solutions. One, Universities should directly contract with Pearson to buy licensing for the materials at a discount. The bookstore prices are often slightly higher than just buying from other sources (Pearson or Amazon) directly. (Students on Pell Grants need to buy from the bookstore though to be reimbursed.)

A second option is simply to pay someone to create open access materials to swap out. Universities often have an option for taking a sabbatical to write a text book. I am pretty sure GSU could throw 30k at an adjunct and they would write just as high (if not higher) quality material. For basic material like that, the current LLM tools could help speed the process by quite a bit.

For these types of textbooks, professors use them because they are convenient, so if a lower cost option were available that met the same needs, I am pretty sure you could convince the math department to have those materials as the standard. If we go to page two of the dashboard though, we see some new types of books pop up:

You may wonder, what is Conley Smith Publishing? It happens to be an idiosyncratic self publishing platform. Look, I have a self published book as well, but having 800 business students a semester buy your self published $100 using excel book, that is just a racket. And it is a racket that when I give that example to friends almost everyone has experienced in their college career.

There is no solution to the latter professors ripping off their students. It is not illegal as far as I’m aware. I am just guessing at the margins, that business prof is maybe making $30k bonus a semester forcing their students to buy their textbook. Unlike the academic textbook scenario, this individual will not swap out with materials, even if the alternative materials are higher quality.

To solve the issue will take senior administration in universities caring that professors are gouging their (mostly low income) students and put a stop to it.

This is not a unique problem to GSU, this is a problem at all universities. Universities could aim to make low/no-cost, and use that as advertisement. This should be particularly effective advertisement for low income serving universities.

If you are interested in a similar analysis for your own university, feel free to get in touch with either myself or Scott. We would like to expand our cost saving projects beyond GSU.

Leave a comment
by Andy Wheeler on February 6, 2025  •  Permalink
Posted in data science, scholarly
Tagged , ,

Posted by Andy Wheeler on February 6, 2025

https://andrewpwheeler.com/2025/02/06/how-much-do-students-pay-for-textbooks-at-gsu/

Bitcoin, Ethereum, and Axon

In 2022, I did a post on the cointegration between Bitcoin, Ethereum, Gold, and the S&P 500. I have had a few conversations about Bitcoin recently with friends and family, so figured it would be worth updating that post.

Also had a discussion with a friend about Axon last week, and when talking about stock I said “what is it at $200” and his reply was “It is close to $700”. So throwing them in the mix as well.

Here is the same indexed between 0/1 chart, so you can see all the different investments appear to be pretty correlated. Since mid 2022 all have been on a fairly consistent upward trajectory.

Now the way this chart works is y = x - min(x)/(max(x) - min(x), where x is the closing price (sampled every Friday). This hides the variation, plotting the closing prices on the logged scale better shows how volatile the different stocks are. So the S&P is a steady march, Gold has been quite consistent, the others not so much.

And a final way to show the data is to index to a start point. Here my initial post was in February 2022, so I start from there and Closing/Closing2_11_2022. So a value of 2 means it doubled from its start point, 3 tripled etc.

I was prompted to look at Ethereum back then due to the popularity of NFTs. I decided not to invest, and if looking as of last Friday, I would be at about the exact same position as I would have been with the S&P. But I would have been under for almost two years. And it would have been almost the exact same story for Bitcoin over the same period, only with the more recent increase would I have really beat the more tame investments from Bitcoin. Axon though!

With the talks about US government investment in Bitcoin (see Bryan’s Clear Value Tax video on the subject), given that traditional index funds do not cover them, I will have to consider it more seriously. (I have had 401k’s that included Gold, I presume they have not expanded to include crypto though.) They are definitely more speculative than investing in index funds, but that can be a good thing if you know that going in.

Given the NFT fad passed with Eth, I wondered if it peaked (it was already falling from a local high of 3k by that time in February 2022 I initially wrote that post). But just a few hours before writing this post, BlackRock and Fidelity purchased a big chunk, so Eth should likely continue to climb at least in the short term and is not entirely pegged to the popularity of NFTs.

The original post I wrote about cointegration analysis, which is really only useful for very short term. Thinking about more long term investments, Bitcoin is harder to peg. The Clear Value Tax video shows Powell comparing Bitcoin to Gold, which I think is probably a good way to view it. So I think you can legitimately view it as a hedge against more traditional investments at this point (and ditto for Ethereum – does CoinBase have an index fund for coins?).

Now when evaluating specific companies, whether something is a good investment is more about whether you think the company itself is on a good trajectory. As disclosure I don’t have any direct financial ties to Axon, nor have I invested in them directly beyond if my 401k has them in the portfolio. I think Axon’s rise is legit and not a fad.

So Axon currently has the dominant market share for body worn cameras and conducted energy devices. Body worn cameras I think are likely to expand out into other areas beyond police and corrections officers. There are some news articles for store workers, I suspect medical workers and teachers though are bigger expanding markets in the future. Motorola stock is not doing too shabby over this time period as well, so they may be able to capture some of that same market as well.

I am not as in the know about drones, but I presume their hardware experience for BWC and Taser make them well positioned to expand out drone capabilities.

I am not sure what prompted the most recent rise mid 2024 for Axon. I wondered if it lined up with Draft One (the generative AI to help write reports), but it appears slightly after that. Their software products I am not as bullish about offhand. I think Draft One has very good potential, although see a recent article by Ian Adams and colleagues showing it does not decrease time writing reports.

But given they have such strong market share in other areas (like BWC) they have already established sales relationships. And if they can figure out BWC they have the technical capabilities to figure out the data software stuff, like RMS as well. Basically all the different “analytic” companies have no moat – Axon could hire me (or various other data scientists) to build those same analytic software programs directly for Axon’s different current software products.

1 Comment
by Andy Wheeler on December 12, 2024  •  Permalink
Posted in data science, Data Visualization
Tagged , , ,

Posted by Andy Wheeler on December 12, 2024

https://andrewpwheeler.com/2024/12/12/bitcoin-ethereum-and-axon/

Identifying excess rounding

Given the hubbub about Blue Cross and paying for anesthesiologists, there was an interesting paper making the rounds, Comparison of Anesthesia Times and Billing Patterns by Anesthesia Practitioners. (Shared via Crémieux.)

Most medical claims are billed via what are call CPT codes (Current Procedural Terminology). So if you go to the ER, you will get billed for a code 99281 to 99285. The final digit encodes different levels of complexity for the case, with 5’s being more complex. It was news to me that anesthesiologists actually bill for time directly, but the above linked paper showed (pretty plainly) that there is strong evidence they round up to every 5 minutes.

Now the paper just selected the anesthesiologists that have the highest proportion of billed times ending in 5’s. Here I will show a better way to flag specific problematic anesthesiologists (using repeated binomial tests and false discovery rate corrections).

Here I simulate 1,000 doctors, and select 40 of them to be bad, and those 40 round all of their claims up to the nearest 5 minute mark. Whereas the other docs are just billing the time as is. And they have varying total number of claims, between 100 and 500.

import numpy as np
from scipy.stats import gamma,uniform
import matplotlib.pyplot as plt
import pandas as pd
from scipy.stats import binomtest,false_discovery_control

np.random.seed(10)

docs = 1000
# pick a random set 
bad_docs = np.random.choice(np.arange(docs),40,replace=False)
doci = []
claims = []

for i in range(docs):
    totn = int(np.ceil(uniform.rvs(100,500))) # number of claims
    doci += [i]*totn
    g = gamma.rvs(6,scale=12,size=totn)
    if i in bad_docs:
        g = np.ceil(g/5)*5
    else:
        g = g.round()
    claims += g.tolist()

dat = pd.DataFrame(zip(doci,claims),columns=['Doc','Time'])

# Histogram
fig, ax = plt.subplots(figsize=(8,4))
dat['Time'].hist(bins=range(201),alpha=0.8,color='k',ax=ax)
plt.savefig('Histogram.png',dpi=500,bbox_inches='tight')

You can see my gamma distribution is not as heavy tailed as the JAMA paper, but qualitatively has the same spike traits. Based on this, you can see the spike is relative to the other density, and so that shows in the JAMA paper there are more anesthesiologists rounding at 60 minutes than there are at the other 5 minute intervals.

In this particular example, it would be trivial to spot the bad docs, since they round 100% of the time, and you would only expect around 20% (since billing in minute intervals).

dat['Round'] = dat['Time'] % 5 == 0
dat['N'] = 1
gs = dat.groupby('Doc',as_index=False)[['N','Round']].sum()
gs['Perc'] = gs['Round']/gs['N']
gs.sort_values(by='Perc',ascending=False,ignore_index=True,inplace=True)
gs

# Upper Quantiles
np.quantile(gs['Perc'],[0.75,0.8,0.85,0.9,0.95,0.99])

But you can see a problem with using a top 5% quantile cut off here. Since I only have 4% bad doctors, using that hard cut-off will result in a few false positive flags. My suggested approach is to create a statistical test (Chi-Square, binominal, KS, whatever makes sense for individual doctors). Run the test for each doctor, get the p-value, and then run a false discovery rate correction on the p-values.

The above where doctors round 100% of the time is too easy, so here I simulate 40 doctors will round up 10% to 30% of the time. I also have fewer cases (more cases and more rounding make it much easier to spot).

# Redoing sim, but it is a smaller percentage of the stats are bad
for i in range(docs):
    totn = int(np.ceil(uniform.rvs(100,500))) # number of claims
    doci += [i]*totn
    g = gamma.rvs(6,scale=12,size=totn)
    if i in bad_docs:
        randperc = int(np.round(totn*uniform.rvs(0.1,0.2)))
        badn = np.random.choice(np.arange(totn),randperc,replace=False)
        g[badn] = np.ceil(g[badn]/5)*5
        g = g.round()
    else:
        g = g.round()
    claims += g.tolist()

dat = pd.DataFrame(zip(doci,claims),columns=['Doc','Time'])
dat['Round'] = dat['Time'] % 5 == 0
dat['N'] = 1
gs = dat.groupby('Doc',as_index=False)[['N','Round']].sum()
gs['Perc'] = gs['Round']/gs['N']
gs.sort_values(by='Perc',ascending=False,ignore_index=True,inplace=True)

Now we can apply the binomial test to each doctor, then adjust for false discovery rate.

# Calculating binom test
def bt(x):
    k,n = x.iloc[0],x.iloc[1]
    b = binomtest(k,n,p=0.2,alternative='greater')
    return b.pvalue

gs['p'] = gs[['Round','N']].apply(bt,axis=1)
gs['q'] = false_discovery_control(gs['p'],method='by')

# Captures 28 out of 40 bad docs, no false positives
gs['BadDocs'] = gs['Doc'].isin(bad_docs)
gs[gs['q'] < 0.05]

If you check out the doctors via gs.head(50), you can see that a few of the bad-docs where adjusted to have q-values of 1, but they ended up being low N and in the range you would expect.

While anesthesiologists billing is different, this same approach would be fine for CPT codes that have the 1-5 modifier (you might use a leave one out strategy and a Chi-Square test). Anesthesiologists if they know they will be scrutinized with exact 5 minutes, they will likely adjust and round up, but to not regular numbers. If that is the case, the default distribution will be expected to be uniform on the 0-9 digits (sometimes people use a Benford like test for the trailing digits, this is the same idea). That will be harder to fake.

I don’t have an issue with Blue Cross saying they will only bill for pre-allotted times. But even without making an explicit policy like that, they can identify bad actors and pursue investigations into those problematic anesthesiologists even without making widespread policy changes.

2 Comments
by Andy Wheeler on December 7, 2024  •  Permalink
Posted in Crime Analysis, data science, Data Visualization, Python
Tagged ,

Posted by Andy Wheeler on December 7, 2024

https://andrewpwheeler.com/2024/12/07/identifying-excess-rounding/

Year in Review 2024

Past year in review posts I have made focused on showing blog stats. Writing this in early December, but total views will likely be down this year – I am projecting around 140,000 views in total for this site. But I have over 25k views for the Crime De-Coder site, so it is pretty much the same compared to 2023 combining the two sites.

I do not have a succinct elevator speech to tell people what I am working on. With the Crime De-Coder consulting gig, it can be quite eclectic. That Tukey quote being a statistician you get to play in everyone’s backyard is true. Here is a rundown of the paid work I conducted in the past year.

Evidence Based CompStat: Work with Renee Mitchell and the American Society of Evidence Based Policing on what I call Evidence Based CompStat. This mostly amounts to working directly with police departments (it is more project management than crime analysis) to help them get started with implementing evidence based practices. Reach out if that sounds like something your department would be interested in!

Estimating DV Violence: Work supported by the Council on CJ. I forget exactly the timing of events. This was an idea I had for a different topic (to figure out why stores and official reports of thefts were so misaligned). Alex approached me to help with measuring national level domestic violence trends, and I pitched this idea (use local NIBRS data and NCVS to get better local estimates).

Premises Liability: I don’t typically talk about ongoing cases, but you can see a rundown of some of the work I have done in the past. It is mostly using the same stats I used as a crime analyst, but in reference to civil litigation cases.

Patrol Workload Analysis: I would break workload analysis for PDs down into two categories, advanced stats and CALEA reports. I had one PD interested in the simpler CALEA reporting requirement (which I can do for quite a bit cheaper than the other main consulting firm that offers these services).

Kansas City Python Training: Went out to Kansas City for a few days to train their analysts up in using python for Focused Deterrence. If you think the agenda in the pic below looks cool get in touch, I would love to do more of these sessions with PDs. I make it custom for the PD based on your needs, so if you want “python and ArcGIS”, or “predictive models” or whatever, I will modify the material to go over those advanced applications. I have also been pitching the same idea (short courses) for PhD programs. (So many posers in private sector data science, I want more social science PhDs with stronger tech skills!)

Patterson Opioid Outreach: Statistical consulting with Eric Piza and Kevin Wolff on a street outreach intervention intended to reduce opioid overdose in Patterson New Jersey. I don’t have a paper to share for that at the moment, but I used some of the same synthetic control in python code I developed.

Bookstore prices: Work with Scott Jacques, supported by some internal GSU money. Involves scraping course and bookstore data to identify the courses that students spend the most on textbooks. Ultimate goal in mind is to either purchase those books as unlimited epubs (to save the students money), or encourage professors to adopt better open source materials. It is a crazy amount of money students pour into textbooks. Several courses at GSU students cumulatively spend over $100k on course materials per semester. (And since GSU has a large proportion of Pell grant recipients, it means the federal government subsidizes over half of that cost.)

General Statistical Consulting: I do smaller stat consulting contracts on occasion as well. I have an ongoing contract to help with Pam Metzger’s group at the SMU Deason center. Did some small work for AH Datalytics on behind the scenes algorithms to identify anomalous reporting for the real time crime index. I have several times in my career consulted on totally different domains as well, this year had a contract on calculating regression spline curves for some external brain measures.

Data Science Book: And last (that I remember), I published Data Science for Crime Analysis with Python. I still have not gotten my 100 sales I would consider it a success – so if you have not bought a copy go do that right now. (Coupon code APWBLOG will get you $10 off for the next few weeks, either the epub or the paperback.)

Sometimes this seems like I am more successful than I am. I have stopped counting the smaller cold pitches I make (I should be more aggressive with folks, but most of this work is people reaching out to me). But in terms of larger grant proposals or RFPs in that past year, I have submitted quite a few (7 in total) and have landed none of them to date! Submitted a big one to support surveys that myself and Gio won the NIJ competition on for place based surveys to NIJ in their follow up survey solicitation, and it was turned down for example. So it goes.

In addition to the paid work, I still on occasion publish peer reviewed articles. (I need to be careful with my time though.) I published a paper with Kim Rossmo on measuring the buffer zone in journey to crime data. I also published the work on measuring domestic violence supported by the Council on CJ with Alex Piquero.

I took the day gig in Data Science at the end of 2019. Citations are often used as a measure of a scholars influence on the field – they are crazy slow though.

I had 208 citations by the end of 2019, I now have over 1300. Of the 1100 post academia, only a very small number are from articles I wrote after I left (less than 40 total citations). A handful for the NIJ recidivism competition paper (with Gio), and a few for this Covid and shootings paper in Buffalo. The rest of the papers that have a post 2019 publishing date were entirely written before I left academia.

Always happy to chat with folks on teaming up on papers, but it is hard to take the time to work on a paper for free if I have other paid work at the moment. One of the things I need to do to grow the business is to get some more regular work. So if you have a group (academic, think tank, public sector) that is interested in part time (or fractional I guess is what the cool kids are calling it these days), I would love to chat and see if I could help your group out.

1 Comment
by Andy Wheeler on December 3, 2024  •  Permalink
Posted in Crime Analysis, Criminal Justice, data science, Personal Productivity, website, writing
Tagged ,

Posted by Andy Wheeler on December 3, 2024

https://andrewpwheeler.com/2024/12/03/year-in-review-2024/

Question Sets and All Paths

I was nerd-sniped with a question at work the other day. The set up was like this, imagine a survey where all of the questions have yes-no answers. Some of the answers are terminating, so if you answer No to a particular question, the questions stop. But some if you reach that part of the tree will keep going.

The ask was a formula to articulate the total number of potential unique answer sets. Which I was not able to figure out, I did however write python code to generate all of the unique sets. So here is that function, where I create a networkx directed graph in a particular format, and then find all of the potential start to end paths in that graph. You just add a dummy begin node, and end nodes at the terminating locations and the final question. You then just search for all of the paths from the begin to end nodes.

import networkx as nx

def question_paths(totq,term):
    '''
    totq -- positive integer, total number of questions
    term -- list of integers, where ints are questions that terminate
    '''
    nodes = [f'Q{i}{r}' for i in range(totq) for r in ['N','Y']]
    edges = []
    for i in range(totq-1):
        edges.append([f'Q{i}Y',f'Q{i+1}N'])
        edges.append([f'Q{i}Y',f'Q{i+1}Y'])
        if i not in term:
            edges.append([f'Q{i}N',f'Q{i+1}N'])
            edges.append([f'Q{i}N',f'Q{i+1}Y'])
    # adding in begin/end
    nodes += ['Begin','End']
    edges += [['Begin','Q0N'],['Begin','Q0Y']]
    for t in term:
        edges.append([f'Q{t}N','End'])
    edges += [[f'Q{totq-1}N','End'],[f'Q{totq-1}Y','End']]
    # Now making graph
    G = nx.DiGraph()
    G.add_nodes_from(nodes)
    G.add_edges_from(edges)
    # Getting all paths
    paths = []
    for p in nx.all_simple_paths(G,source='Begin',target='End'):
        nicer = [v[-1] for v in p[1:-1]]
        paths.append(nicer)
    return paths

And now we can check out where all paths are terminating, you only get further up the tree if you answer all yes for prior questions.

>>> question_paths(3,[0,1,2])
[['N'],
 ['Y', 'N'],
 ['Y', 'Y', 'N'],
 ['Y', 'Y', 'Y']]

So in that scenario we have four potential answer sets. For all binary, we have the usual 2^3 number of paths:

>>> question_paths(3,[])
[['N', 'N', 'N'],
 ['N', 'N', 'Y'],
 ['N', 'Y', 'N'],
 ['N', 'Y', 'Y'],
 ['Y', 'N', 'N'],
 ['Y', 'N', 'Y'],
 ['Y', 'Y', 'N'],
 ['Y', 'Y', 'Y']]

And then you can do a mixture, here just question 2 terminates, but 1/3 are binary:

>>> question_paths(3,[1])
[['N', 'N'],
 ['N', 'Y', 'N'],
 ['N', 'Y', 'Y'],
 ['Y', 'N'],
 ['Y', 'Y', 'N'],
 ['Y', 'Y', 'Y']]

One of the things doing this exercise taught me is that it matters where the terminating nodes are in the tree. Earlier terminating nodes results in fewer potential paths.

# can different depending on where the terminators are
len(question_paths(10,[1,5])) # 274
len(question_paths(10,[6,8])) # 448
len(question_paths(10,[0,1])) # 258
len(question_paths(10,[8,9])) # 768

So the best in terms of a formula for the total number of paths I could figure out was 2^(non-terminating questions) <= paths <= 2^(questions) (which is not a very good bound!) I was trying to figure out a product formula but was unable (any suggestions let me know!)

This reminds me of a bit of product advice from Jennifer Pahlka, you should have eliminating questions earlier in the form. So instead of filling out 100 questions and at the end being denied for TANF, you ask questions that are the most likely to eliminate people first.

It works the same for total complexity of your application. So asking the terminating questions earlier reduces the potential number of permutations you ultimately have in your data as well. Good for the user and good for the developers.

2 Comments
by Andy Wheeler on November 2, 2024  •  Permalink
Posted in data science, Python
Tagged ,

Posted by Andy Wheeler on November 2, 2024

https://andrewpwheeler.com/2024/11/02/question-sets-and-all-paths/

Suits, Money Laundering, and Linear Programming

Currently watching Suits with the family, and an interesting little puzzle came up in the show. In Season 1, episode 8, there is a situation with money laundering from one account to many smaller accounts.

So the set up is something like “transfer out 100 million”, and then you have some number of smaller accounts that sum to 100 million.

When Lewis brought this up in the show, and Mike had a stack of papers to go through to identify the smaller accounts that summed up to the larger account, my son pointed out it would be too difficult to go through all of the permutations to figure it out. The total number of permutations would be something like:

N = Sum(b(A choose k) for 1 to k)

Where A is the total number of accounts to look over, and k is the max number of potential accounts the money was spread around. b(A choose k) is my cheeky text format for binomial coefficients.

So this will be a very large number. 10,000 choose 4 for example is 416 416 712 497 500 – over 416 trillion. You still need to add in b(10,000 choose 2) + b(10,000 choose 3) + ….. b(10,000 choose k). With even a small number of accounts, the number of potential combinations will be very large.

But, I said in response to my son I could write a linear program to find them. And this is what this post shows, but doing so made me realize there are likely too many permutations in real data to make this a feasible approach in conducting fraud investigations. You will have many random permutations that add up to the same amount. (I figured some would happen, but it appears to me many happen in random data.)

(Note I have not been involved with any financial fraud examinations in my time as an analyst, I would like to get a CFA time permitting, and if you want to work on projects let me know! So all that said, I cannot speak to the veracity of whether this is a real thing people do in fraud examinations.)

So here I use python and the pulp library (and its default open source CBC solver) to show how to write a linear program to pick bundles of accounts that add up to the right amount. First I simulate some lognormal data, and choose 4 accounts at random.

import pulp
import numpy as np

# random values
np.random.seed(10)
simn = 10000
x = np.round(np.exp(np.random.normal(loc=5.8,scale=1.2,size=simn)),2)

# randomly pick 4
slot = np.arange(0,simn)
choice = np.random.choice(slot,size=4,replace=False)
tot = x[choice].sum()
print(tot,choice)

So we are expecting the answer to be accounts 2756 5623 5255 873, and the solution adds up to 1604.51. So the linear program is pretty simple.

# make a linear program
P = pulp.LpProblem("Bundle",pulp.LpMinimize)
D = pulp.LpVariable.dicts("D",slot,cat='Binary')

# objective selects smallest group
P += pulp.lpSum(D)

# Constraint, equals the total
P += pulp.lpSum(D[i]*x[i] for i in range(simn)) == tot

# Solving with CBC
P.solve()

# Getting the solution
res = []
for i in range(simn):
    if D[i].varValue == 1:
        res.append(i)

In practice you will want to be careful with floating point (later I convert the account values to ints, another way though is instead of the equal constraint, make it <= tot + eps and >= tot - eps, where eps = 0.001. But for the CBC solver this ups the time to solve by quite a large amount.)

You could have an empty objective function (and I don’t know the SAT solvers as much, I am sure you could use them to find feasible solutions). But here I have the solution by the minimal number of accounts to look for.

So here is the solution, as you can see it does not find our expected four accounts, but two accounts that also add up to 1604.51.

You can see it solved it quite fast though, so maybe we can just go through all the potential feasible solutions. I like making a python class for this, that contains the “elimination” constraints to prevent that same solution from coming up again.

# Doing elimination to see if we ever get the right set
class Bundle:
    def __init__(self,x,tot,max_bundle=10):
        self.values = (x*100).astype('int')
        self.totn = len(x)
        self.tot = int(round(tot*100))
        self.pool = []
        self.max_bundle = max_bundle
        self.prob_init()
    def prob_init(self):
        P = pulp.LpProblem("Bundle",pulp.LpMinimize)
        D = pulp.LpVariable.dicts("D",list(range(self.totn)),cat='Binary')
        # objective selects smallest group
        P += pulp.lpSum(D)
        # Constraint, equals the total
        P += pulp.lpSum(D[i]*self.values[i] for i in range(simn)) == self.tot
        # Max bundle size constraint
        P += pulp.lpSum(D) <= self.max_bundle
        self.prob = P
        self.d = D
    def getsol(self,solver=pulp.PULP_CBC_CMD,solv_kwargs={'msg':False}):
        self.prob.solve(solver(**solv_kwargs))
        if self.prob.sol_status != -1:
            res = []
            for i in range(self.totn):
                if self.d[i].varValue == 1:
                    res.append(i)
            res.sort()
            self.pool.append(res)
            # add in elimination constraints
            self.prob += pulp.lpSum(self.d[r] for r in res) <= len(res)-1
            return res
        else:
            return -1
    def loop_sol(self,n,solver=pulp.PULP_CBC_CMD,solv_kwargs={'msg':False}):
        for i in range(n):
            rf = self.getsol(solver,solv_kwargs)
            if rf != 1:
                print(f'Solution {i}:{rf} sum: {self.values[rf].sum()}')
            if rf == -1:
                print(f'No more solutions at {i}')
                break

And now we can run through and see if the correct solution comes up:

be = Bundle(x=x,tot=tot)
be.loop_sol(n=10)

Uh oh – I do not feel like seeing how many potential two solutions there are in the data (it is small enough I could just enumerate those directly). So lets put a constraint in that makes it so it needs to be four specific accounts that add up to the correct amount.

# add in constraint it needs to be 4 selected
choice.sort()
be.prob += pulp.lpSum(be.d) == 4
print(f'Should be {choice}')
be.loop_sol(n=10)

And still no dice. I could let this chug away all night and probably come up with the set I expected at some point. But if you have so many false positives, this will not be very useful from an investigative standpoint. So you would ultimately need more constraints.

Lets say our example the cumulative total will be quite large. Will that help limit the potential feasible solutions?

# Select out a sample of high
sel2 = np.random.choice(x[x > 5000],size=4,replace=False)
tot2 = sel2.sum()
ch2 = np.arange(simn)[np.isin(x,sel2)]
ch2.sort()
print(tot2,ch2)

be2 = Bundle(x=x,tot=tot2)
be2.loop_sol(n=20)

There are still too many potential feasible solutions. I thought maybe a problem with real data is that you will have fixed numbers, say you are looking at transactions, and you have specific $9.99 transactions. If one of those common transactions results in a total sum, it will just be replicated in the solution over and over again. I figured with random data the sum would still be quite unique, but I was wrong for that.

So I was right in that I could write a linear program to find the solution. I was wrong that there would only be one solution!

Leave a comment
by Andy Wheeler on October 28, 2024  •  Permalink
Posted in Crime Analysis, data science, Python
Tagged ,

Posted by Andy Wheeler on October 28, 2024

https://andrewpwheeler.com/2024/10/28/suits-money-laundering-and-linear-programming/

GenAI is not a serious solution to California’s homeless problem

This is a rejected op-ed (or at least none of the major papers in California I sent it to bothered to respond and say no-thanks, it could be none of them even looked at it). Might as well post it on personal blog and have a few hundred folks read it.

Recently Gov. Newsom released a letter of interest (LOI) for different tech companies to propose how the state could use GenAI (generative artificial intelligence) to help with California’s homeless problem. The rise in homelessness is a major concern, not only for Californian’s but individuals across the US. That said, the proposal is superficial and likely to be a waste of time.

A simple description of GenAI, for those not aware, are tools to ask the machine questions in text and get a response. So you can ask ChatGPT (a currently popular GenAI tool) something like “how can I write a python function to add two numbers together” and it will dutifully respond with computer code (python is a computer programming language) that answers your question.

As someone who writes code for a living, this is useful, but not magic. Think of it more akin to auto-complete on your phone than something truly intelligent. The stated goals of Newsom’s LOI are either mostly trivial without the help of GenAI, or are hopeless and could never be addressed with GenAI.

For the first stated goal, “connecting people to treatment by better identifying available shelter and treatment beds, with GenAI solutions for a portable tool that the local jurisdictions can use for real-time access to treatment and shelter bed availability”. This is simply describing a database — one could mandate state funded treatment providers to provide this information on a daily basis. The technology infrastructure to accomplish this is not much more complex than making a website. Mandating treatment providers report that information accurately and on a timely basis is the hardest part.

For the second stated goal, “Creating housing with more data and accountability by creating clearer insights into local permitting and development decisions”. Permitting decisions are dictated by the state as well as local ordinances. GenAI solutions will not uncover any suggested solution that most Californian’s don’t already know — housing is too expensive and not enough is being built. This is in part due to the regulatory structure, as well as local zoning opposition for particular projects. GenAI cannot change the state laws.

For the last stated goal of the program, “Supporting the state budget by helping state budget analysts with faster and more efficient policy”. Helping analysts generate results faster is potentially something GenAI can help with, more efficient policy is not. I do not doubt the state analysts can use GenAI solutions to help them write code (the same as I do now). But getting that budget analysis one day quicker will not solve any substantive homeless problem.

I hate to be the bearer of bad news, but there are no easy answers to solve California’s homeless crisis. If a machine could spit out trivial solutions to solve homelessness in a text message, like the Wizard of Oz gifting the Scarecrow brains, it would not be a problem to begin with.

Instead of asking for ambiguous GenAI solutions, the state would be better off thinking more seriously about how they can accomplish those specific tasks mentioned in the LOI. If California actually wants to make a database of treatment availability, that is something they could do right now with their own internal capacity.

Solutions to homelessness are not going to miraculously spew from a GenAI oracle, they are going to come from real people accomplishing specific goals.

If folks are reading this, check out my personal consulting firm, Crime De-Coder. I have experience building real applications. Most of the AI stuff on the market now is pure snake oil, so better to articulate what you specifically want and see if someone can help build that.

Crime De-Coder consulting

Leave a comment
by Andy Wheeler on September 30, 2024  •  Permalink
Posted in data science, scholarly, writing
Tagged ,

Posted by Andy Wheeler on September 30, 2024

https://andrewpwheeler.com/2024/09/30/genai-is-not-a-serious-solution-to-californias-homeless-problem/

« Older Posts
Newer Posts »