In PROC GLMSELECT with SELECTION=LASSO, the PATH= option is specified. What does the solution path represent?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — the LASSO solution path is the sequence of coefficient estimates as lambda varies from large to small. At maximum lambda, all coefficients are 0; as lambda decreases, variables enter the model one by one.
Full Explanation
The correct answer is the sequence of coefficient estimates as the regularization parameter lambda varies from large (all coefficients 0) to small (OLS solution), because the LASSO solution path (also called the regularization path) shows how each coefficient changes as lambda varies. At maximum lambda, all coefficients are 0; as lambda decreases, variables enter the model one by one. The PATH= option saves these coefficient estimates at each lambda value to a dataset. The stepwise algorithm sequence is wrong because stepwise visits specific model configurations, not a continuous regularization path. The directory path where PROC GLMSELECT saves coefficient estimates is wrong because PATH= specifies an output dataset name, not a filesystem directory. The cross-validation fold sequence is wrong because cross-validation fold sequences are not what PATH= represents.