Date created: 2022-07-19
(I don’t understand this super well). In Lasso regularization, there is downward pressure on factors in a model, driving them towards zero. Only the truly strong factors will have a non-zero effect and remain.
You can transform an overfitted model into a simpler, robust formula with only the most important factors remaining.
- Link to website, bibtex from Zotero or note with book/blog/etc summary