Re: Minor Problems With Ridge Regression

Posted by Jon K Peck on Aug 30, 2011; 5:00pm
URL: http://spssx-discussion.165.s1.nabble.com/Minor-Problems-With-Ridge-Regression-tp4750732p4750859.html

I don't know what the invalid character issue is, but Ridge Regression is available directly in the CATREG command (Analyze>Regression>Optimal Scaling in the menus).



Jon Peck (no "h")
Senior Software Engineer, IBM
[hidden email]
new phone: [phone]




From:        Zachary Feinstein <[hidden email]>
To:        [hidden email]
Date:        08/30/2011 10:26 AM
Subject:        [SPSSX-L] Minor Problems With Ridge Regression
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

I am using SPSS V19.0.0.  I have the need to run the Ridge Regression syntax.  I copied the syntax file to a C:\temp2 directory on my machine and I changed the syntax a slight bit to place the rr__tmp1.sav and rr_tmp2.sav files into the temp2 directory as well.  I then run the following:
 
INCLUDE 'C:\temp2\Ridge regression.sps'.
RIDGEREG DEP = dv1
/ENTER = zqtp1 TO zqtp14 zholder
/START = 0
/STOP = 1
/INC = 0.05
 /DEBUG = 'Y'.
 
At the very end if I open up my rr_tmp2.sav file (in my temp2 directory) and it looks and seems to be providing the proper ridge regression coefficients for the various k-levels.  But I received errors or warnings such as the following:
 
Warning # 206 on line 6049 in column 12. Text: \
An invalid character has been found on a command.
This one repeats itself a few times.
 
Error # 11187
The specified report is too wide for the page width. The page width is
controlled by the MARGINS(l,r) phrase on the FORMAT subcommand. The first
number represents the left margin and must be greater than or equal to the
system margin or an *, which signifies the system margin. The second number
represents the right margin and must be less than or equal to 255. When it is
an * or does not appear, it is set equal to the system margin. When MARGINS
is not specified, the system margins are used.
Execution of this command stops.
 
Should I be concerned at all about the quality of my Ridge Regression coefficients?  What may I do to get rid of these warnings and errors?
 
Thank you very much.  I look forward to your responses.  And I apologize if anyone on the list had previously received SPAM e-mail from me.  Someone infiltrated my e-mail account.  This is the first time it has ever happened to me since e-mail was created- and I have been a member of the SPSS list for practically longer than I can remember.
 
Regards,
 
Zachary
zsfeinstein@...
[phone]