How the Lasso Thinks

Basics 1 of 5 · Card 1 of 27

Welcome

Welcome. This is a short, hands-on tour of how a popular kind of regression decides which features to use. All you need is a basic familiarity with regression.

If "lasso" or "elastic net" are new to you, they're simply common forms of regression that penalize, or regularize, the coefficients: they shrink the model, often setting some coefficients exactly to zero, which can help it generalize to new data.

Rather than the math, you'll watch what they do: which features they pick, in what order, and why. It takes about 15 minutes: the basics, a lab of small cases, real data, and takeaways.

Optional background, if you want it:

Hastie, Tibshirani & Wainwright, Statistical Learning with Sparsity
Terence Parr, explained.ai