I'm happy to release free, online textbooks covering the two courses I taught at the NICAR26 data journalism conference put on by Investigative Reporters and Editors last week in Indianapolis.
Everything we did in class is scripted out with setup instructions, code snippets and screenshots.
First LLM Classifier, taught with Andrew Briz, will show you how journalists use large-language models to organize and analyze mammoth datasets.
Our script will teach you how to submit dozens of prompts at once as part of an automated routine, and then evaluate and improve results using a rigorous, scientific approach. This year's edition is a significant improvement over last year's experimental run.
You can find it at https://palewi.re/docs/first-llm-classifier/.
First PMTiles Map, taught with Jon Schleuss, will show how to display a massive dataset on an interactive map using Protomaps and MapLibre.
It will teach you how to convert heavy shapefiles to the compact PMTiles format, and then display the tiles on an interactive map using a data-driven style. We had a fun debut at the conference, and I'm excited to hear what people think of this technique.
You can find it at https://palewi.re/docs/first-pmtiles-map/.