A ground-up walk through the mathematics of linear regression — least squares, variance partitioning, significance testing, maximum likelihood, and Bayesian regression — built from James V. Stone’s “Linear Regression with Python” and reframed through the lens of a working data scientist.
Linear regression is usually the first model anyone learns, and often the last one anyone really thinks about. It’s easy to fit a line with one call to sklearn and move on. But every time I’ve had to explain why a fitted line should be trusted to myself — I’ve realised how much of the intuition underneath gets skipped.
A ground-up walkthrough of the event study method from MacKinlay (1997) — covering the intuition, the mathematics, and a deliberate translation into modern data science language. If you work with time series and want to measure the effect of an intervention, this toolkit is more useful than you might expect.
There is a method in empirical economics that has been quietly doing what data scientists now call causal inference since the 1930s. It is called the event study. The canonical reference is A. Craig MacKinlay’s 1997 paper in the Journal of Economic Literature, and while the paper frames everything around stock prices and earnings announcements, the underlying machinery is general enough to apply to almost any time series problem where you want to measure the effect of a well-defined event.
A deep dive into how the board game Turing Machine works — from its punch card mechanism and Verification card design to puzzle generation and the genuine computer science concepts playing out on your table.
The name alone caught my attention. Turing Machine — a board game named after one of the most important theoretical constructs in computer science. I came across reviews describing how it used physical punch cards as a coding device, and I was curious enough to buy a copy and find out whether the connection was more than just a marketing nod.