Regression Shrinkage and Selection via the Lasso

Details

Photo

—

Name

—

Authors

Robert Tibshirani

Year

1996

Field

Statistics / Machine Learning

Why it matters

Introduces LASSO, a way to pick out the truly important variables while keeping models simple and avoiding overfitting.

My take

If you’ve ever tried to jam a hundred tracking stats into one model and gotten mush back, this is the paper that teaches you to cut dead weight like a GM on deadline day.