Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: 95% CI's for Stepwise Regression (-swboot-)


From   Nick Cox <njcoxstata@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: 95% CI's for Stepwise Regression (-swboot-)
Date   Tue, 27 Nov 2012 08:43:31 +0000

On -swboot-: It is from SSC. Please read the Statalist FAQ on proper
referencing of user-written commands.

On stepwise: See

FAQ     . . . . . . . . . . . . . . . . . .  Problems with stepwise regression
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W. Sribney
        5/98    What are some of the problems with stepwise regression?

http://www.stata.com/support/faqs/statistics/stepwise-regression-problems/

and Harrell, F. 2001. Regression modeling strategies. New York: Springer.

You can easily get CIs for RMSE using plain -bootstrap-.

. sysuse auto
. regress mpg weight
. bootstrap rmse=e(rmse) _b , reps(1000): regress mpg weight

Nick

On Tue, Nov 27, 2012 at 6:47 AM, rasool.bux <rasool.bux@aku.edu> wrote:

> Thanks for reply. But I want to compute 95% CI's for RMSE of each sample if I use -bsample- in a loop for my data even for regression.

William Buchanan

> If you've searched through the Statalist Archives at all, you'll have noticed that in general many people do not support the use of stepwise regression for multiple valid reasons. Why not just fit a model to your data that is supported by your theory with a regular linear regression model instead?

On Nov 23, 2012, at 4:32, "rasool.bux" <rasool.bux@aku.edu> wrote:

>> How could I get 95%CI's for stepwise regression model. I found -swboot- but it is not giving any estimates. Is there any simple way to get 95%CI's for sw regression.  I am doing the following procedures to get CI but I am not sure that is this a correct way or not.
>>
>> Procedure #1
>> ==========
>> 1. I am running sw regression
>> 2. Saving predicted values
>> 3. Running bootstrap with regress using yvariable and yhat variable.
>> 4. This is giving the CI's but the Observed RMSE is different from the original as in #1 and I think it is b/c of degrees of freedom.
>>
>> Prorcedure #2
>> ============
>> 1. I am running sw regression
>> 2. Running bootstrap with regress using yvariable and those variables which are kept by the model.
>> 3. This is giving the same observed RMSE as in #1.
>>
>> I am not clear Which CI's should I report and which method is appropriate and correct. Kindly help me in this regard

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index