Dear all,
For variable selection using penalized methods, it is of crucial importance to select the appropriate tuning parameters through a grid search over many points.
How can i determine the grid points for this parameter, in other words, how determined the possible values (lambda.min and lambda.max) of lambda?
Jhonathan Pedroso Rigal dos Santos
Actually its depends on the sample space of the random variable. For example, if you are tuning a correlation parameter you should create a grid with small spacing values though -1 to 1. If it is a variance component, you can use some empirical estimate only using the data to give you an idea, and them create a grid with some variation across both sides of the empirical estimate, and so on.
I hope to be helpful,
1 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression
More Mohammad Kazemi's questions See All
Similar questions and discussions