One studio a week on a real question: a dataset, a baseline, or a model result. The room argues about what the number means.
Labs on your own clock. You submit the measurement and a short account of where it could be lying.
- 01
Python for a question
Enough language to work with data. The question stays in front.
- 02
What a dataset can answer
Rows, leakage, and the question you were not actually asking.
- 03
Scores that hide
Baselines, class balance, and the measurement behind a percentage.
- 04
Simpler models
When a linear model beats something fashionable.
- 05
Large models, audited
Using a modern model with checks a person can review.
- 06
An explanation someone can use
Writing the result for a person who might act on it.
- 07
A capstone you can defend
One project you can walk through line by line.