2

I have a txt file containing a large set of data. How do I go about using the lasso or the ridge estimator to fit a regression equation?

I got as far as using: gridge

However, I am unsure of what that lambda section is doing. I found it on a website and have no idea what values to put there.

And I don't know how to interpret the output: modified HKB estimator is 5.465433 modifiedL-W estimator is 7.6435664 smallest value of GCV at 3.24

How am I supposed to fit a regression equation using that information?

math11
  • 537
  • 2
  • 6
  • 8
  • You've asked 5 questions and accepted none of the answers. If the answers are no satisfactory you should be making comments to guide the respondents. If they are satisfactory you should upvote and apply checkmarks. – IRTFM Nov 03 '16 at 16:43

1 Answers1

2

Lambda is the strength of the penalty, See this to understand better the effect of lambda. Generally, you select this value by try-and-error or using cross validation procedure.

Both Lasso and Ridge estimation help to reduce the model over fitting by limiting the value of the parameters to be estimated. The main difference between them is the shape of the penalty function.

Lasso can result in a sparse model where some parameters can be exactly zero, while Ridge can lead to parameters with very small value but not exactly zero.

iTech
  • 18,192
  • 4
  • 57
  • 80