Question
Robert Tibshirani coined the name for this statistical method in a 1996 JRSS-B article, where he noted the favorable properties of this method’s use of an L1 penalty. For 10 points each:
[10h] Name this technique, which in the Elastic Net is functionally combined with a similar method that uses an L2 penalty. This technique’s L1 penalty shrinks some coefficients to true zero and performs subset selection.
ANSWER: LASSO [or Least Absolute Shrinkage and Selection Operator]
[10m] This data resampling technique is often used to find a penalty parameter for LASSO. The statistics section of Stack Exchange is named for this technique, which comes in K-fold and Leave-one-out varieties.
ANSWER: cross-validation [or CV; accept K-fold cross-validation; accept Leave-one-out cross-validation or LOOCV; accept Cross Validated; prompt on validate or validation or validating a model]
[10e] LASSO is an alternative to stepwise selection, which often selects variables with this property of a p-value being below a chosen cutoff alpha like .05 (“point zero five”) and hence rejecting a null hypothesis.
ANSWER: statistical significance [or statistically significant; accept Null Hypothesis Significance Testing]
<JF, Other Science: Math>
Summary
| 2023 ARCADIA at UC Berkeley | Premiere | Y | 1 | 30.00 | 100% | 100% | 100% |
| 2023 ARCADIA at Carleton University | Premiere | Y | 3 | 16.67 | 100% | 33% | 33% |
| 2023 ARCADIA at RIT | Premiere | Y | 1 | 30.00 | 100% | 100% | 100% |
| 2023 ARCADIA at WUSTL | Premiere | Y | 3 | 3.33 | 33% | 0% | 0% |
Data
| Stanford A | Berkeley A | 10 | 10 | 10 | 30 |