eli knaap
banner
knaaptime.com
eli knaap
@knaaptime.com
urban planning & public policy prof @ UC Irvine | core dev @ PySAL & QuantEcon | urban social science & spatial data science | open source

https://knaaptime.com
Pinned
there's a ~complete rough draft of Urban Analysis up on my website (as in, code executes all the way through)

knaaptime.com/urban_analysis

still lots of editing to do, but if you're into cities, Python, or spatial analysis, give it a look and let me know what's wrong :)
Reposted by eli knaap
A long-awaited update of federal employment data shows crippling staff reductions in some federal agencies and offices:
www.nytimes.com/interactive/...
220,000 Fewer Workers: How Trump’s Cuts Affected Every Federal Agency (Gift Article)
New data offers the first clear view of the impact of the buyouts and firings.
www.nytimes.com
January 9, 2026 at 9:33 PM
Reposted by eli knaap
Announcing deck.gl-raster: 𝐟𝐮𝐥𝐥𝐲 𝐜𝐥𝐢𝐞𝐧𝐭-𝐬𝐢𝐝𝐞 𝐂𝐎𝐆 𝐫𝐞𝐧𝐝𝐞𝐫𝐢𝐧𝐠. No server required.

1.3 𝐠𝐢𝐠𝐚𝐛𝐲𝐭𝐞 COG, streamed directly into the browser: developmentseed.org/deck.gl-rast...

- GPU-accelerated raster reprojection
- GPU image processing for colormaps, nodata values
- Efficient use of COG overviews
January 7, 2026 at 7:11 PM
news from PySAL world: we have a new package in the ecosystem(!) As of today, `gwlearn` is available in package managers. It provides a suite of tools for geographically weighted machine learning using a sklearn API.

pysal.org/gwlearn/v0.1...

great work by lead author @martinfleischmann.net
gwlearn
pysal.org
January 9, 2026 at 7:09 PM
another sweet port from Kyle! A touch of background on the original, for those interested:

knaaptime.com/urban_analys...
January 7, 2026 at 8:01 PM
Reposted by eli knaap
My favorite #Python package to use is spopt, a library for spatial optimization.

It helps you with:

📊 Facility location planning;
📊 Sales territory design;
📊 Maximizing market share;

And much more! Check it out here:

pysal.org/spopt/
December 27, 2025 at 2:56 PM
Reposted by eli knaap
Is logistic regression "AI"? What about convex optimization? Is LLM AI? What about an image recognition system built without using "feature learning" techniques but done through "feature engineering"?
December 22, 2025 at 5:22 PM
classic.
December 22, 2025 at 11:13 PM
furthermore, i dont care how this game turns out, or how many caleb williams apoligistst there are, malik is a better qb
LaFleur completed outcoached Johnson tonight. Bears absolutely deserve to lose and until they beat Green Bay none of this means anything.
December 21, 2025 at 4:23 AM
tom. it's not called the windy city because of the weather.
December 21, 2025 at 3:42 AM
seen this floating around all day, with exactly this thought every time. If you're prepared to spend a 7th of this much, you have not spent enough time in the midwest.

the Home Alone house has been valued under 3M until 2024 and its *in the chicago metro*

www.zillow.com/homedetails/...
December 21, 2025 at 2:29 AM
among the many distinctions in *geospatial analysis*, it seems like physical geographers use "AI" in the 'traditional' sense of, like, deep nets applied to imagery data. whereas human geographers use it in the LLM sense of, like 'please vibecode my work for me because i asked so nicely'
Open Geospatial Workshop happening at #AGU25 in Room 333 today! I will be teaching interactive mapping and GeoAI from 9:45-11:45 AM.

Registration is required is to attend, but I heard that they will be checking badges at the entrance until 9:30 AM.
December 20, 2025 at 4:06 AM
ima let you guys finish, but i live down the street from a frazier farms
Trying to explain HEB to a Southern Californian "so imagine a Ralph's/Von's but independent, better, and food quality more like TJ's"

"Oh so like a big Sprouts"

This poor sweet baby
December 19, 2025 at 2:48 AM
this, 💯
the tea party was really a seminal shift in how the median congressperson behaved in person

you stopped needing to be a successful schmoozer and started needing to be a totally unreasonable lunatic to win your primaries, and social media has only accelerated this
My theory that as members of congress are increasingly batshit it will increasingly select for even more batshit people who can tolerate the job is doing pretty well
December 19, 2025 at 1:28 AM
it's kinda wild that the virtual line judge is basically never used to check first downs in the nfl (as of yet), but like every single offside call in the premier league is assessed with VAR (football/football)
December 19, 2025 at 1:26 AM
TIL i learned colab comes with pysal 25.7 preinstalled.

and, thats cool.
December 13, 2025 at 7:32 AM
unrelated, im working on a proper 'big data' project with data much larger than memory (so requires lazy/distributed compute) and i guess im the only one on the internet who really dislikes the polars api. maybe just because its unfamiliar but i find it really cumbersome
December 13, 2025 at 2:13 AM
was reading through the pandas3 release notes the other day and noticed they added support for [bodo](docs.bodo.ai/quick_start/...) as an engine in `apply` (already i was unaware you can use numba as an engine)

ive never heard of bodo but it seems *awesome*? it's, like, jit and parallel at once?
GitHub - bodo-ai/Bodo: High Performance Data Processing in Python
High Performance Data Processing in Python. Contribute to bodo-ai/Bodo development by creating an account on GitHub.
github.com
December 13, 2025 at 2:11 AM
Reposted by eli knaap
PACKERS WIN THERE IS HOPE FOR AMERICA
December 8, 2025 at 12:24 AM
caleb sucks! the bears are trash! 😝😝😝😝😝
dont let the commentators lie to you. caleb sucks. the bears are trash. and everyone in the nfc is deeply embarrassed chicago has a number one seed

these are facts
December 8, 2025 at 12:23 AM
Reposted by eli knaap
I would love to see Parsons sack a ref
December 8, 2025 at 12:15 AM
we just not gonna call micah getting held on every play?
December 8, 2025 at 12:16 AM
dont let the commentators lie to you. caleb sucks. the bears are trash. and everyone in the nfc is deeply embarrassed chicago has a number one seed

these are facts
December 7, 2025 at 10:22 PM
man.

@andrew.heiss.phd is [an open source] legend. I'm putting together quarto websites for my winter courses and following his examples is just impossibly useful. Down to the way he parameterizes using variables so content is reusable

(i've also used his hikmah manuscript templates for ages)
December 4, 2025 at 4:11 AM
/end ramble
December 2, 2025 at 4:39 AM
wow.. where did lets-plot.org come from?

...undeniably impressive, especially the the geo-support.
Lets-Plot - Multiplatform Plotting Library Based on the Grammar of Graphics
Lets-Plot - a faithful port of R’s ggplot2 to Python and Kotlin.
lets-plot.org
December 2, 2025 at 4:39 AM