LASSO
Units & Measurements abbreviation · Statistics
| Stands for | Least absolute shrinkage and selection operator |
|---|---|
| Categories | Units & Measurements |
LASSO stands for Least absolute shrinkage and selection operator, a statistics abbreviation.
A regression method that adds a penalty on the sum of absolute coefficient values, shrinking some coefficients to exactly zero and thereby selecting variables. It is useful when there are many candidate predictors.
Example: LASSO kept 12 of the 300 candidate predictors in the final model.
Usage notes
Robert Tibshirani introduced the lasso in 1996. Ridge regression uses a squared penalty that shrinks but never zeroes coefficients, and the elastic net combines both. The penalty strength is usually chosen by cross-validation, as in R's glmnet package.
Related abbreviations
More statistics abbreviations
- ICC – Intraclass correlation coefficient
- iid – Independent and identically distributed
- IQR – Interquartile range
- IV – Instrumental variable
- KDE – Kernel density estimation
- KS – Kolmogorov-Smirnov (test)
- LDA – Linear discriminant analysis
- LLN – Law of large numbers
- M – Mean (APA style)
- MAD – Mean absolute deviation
- MAE – Mean absolute error
- MANOVA – Multivariate analysis of variance