Page 1 of 1

Maximum Group Size for Regression

Posted: Fri Mar 20, 2009 3:26 am
by dieterE28
I want to run a simple regression using a group of series (ca. 1000).

y c GROUP

This results in the error "752 is not a valid index for vector-series-coefficient C". Is there a maximum size for regressors? Or did I make another mistake.

Re: Maximum Group Size for Regression

Posted: Fri Mar 20, 2009 8:07 am
by EViews Gareth
740 is the maximum number of coefficients for an Equation.

Systems will let you have more coefficients, so you could estimate your equation in a System, which should give you the same results as doing it in an equation would.

Be sure to use the Proc->Define System dialog to make specifying the equation simple with your group object, and use a coefficient vector other than C (C is limited to 740 values, but user-defined coefficient vectors can go larger).

Re: Maximum Group Size for Regression

Posted: Fri Mar 20, 2009 8:53 am
by dieterE28
Thanks. However, I have the same Problem now with the system. I cannot put all series in it. If I do it with copy and paste, I get the same Error message (752 is not...).

Re: Maximum Group Size for Regression

Posted: Fri Mar 20, 2009 9:00 am
by EViews Gareth
Did you change the coefficient vector to something other than C?

Re: Maximum Group Size for Regression

Posted: Tue Feb 27, 2024 1:29 pm
by nmark@nd.edu
Can you give a little example about how to change the coefficient vector to something other than c?
I am having the same problem in a system. I have more than 740 coefficients, but replacing c(1)-c(740) with something like g(1)-g(740) doesn't work.

Re: Maximum Group Size for Regression

Posted: Tue Feb 27, 2024 1:56 pm
by EViews Gareth
Create a new coefficient vector (not a vector, but a coef) called G, then use g instead of c