Lasso
LASSO
- Date
- 1996 · confidence high
- Place
- Toronto, Ontario, Canada
- Credited to
- Robert Tibshirani
- Field
- Knowledge & symbols
What it is
A regression method that performs variable selection and regularization at once through an L1 penalty, improving the interpretability and accuracy of linear models; formulated by Robert Tibshirani in 1996.
What it led to
- Extension to Elastic Net and other combined regularizations
- Widespread use in machine learning and high-dimensional genomics
- Automatic variable selection in models with many predictors