Timeline for What problem do shrinkage methods solve?
Current License: CC BY-SA 3.0
2 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 21, 2020 at 10:00 | comment | added | Jonas Striaukas | It is worth pointing out that we should know before hand what we want to achieve. That is, if we want to do prediction, i.e. compute E[Y|X], we definitely care about bias-variance trade-off and want to minimize it (so exactly what you mention). However, if the goal is to do inference, then we care if the estimator is biased or not. In those cases, we would used debiased (LASSO or other) estimators, where you fit the penalized regression, but also correct for the bias that is introduced due to penalization. | |
| Dec 28, 2011 at 10:44 | history | answered | Dikran Marsupial | CC BY-SA 3.0 |