Page 1 of 1

Constrained least square and optimization

Posted: Tue Oct 27, 2015 2:38 pm
by Roman
Hello,

I have EViews 9. I would have 2 questions about constrained least square estimation:
1. Let's say I want to estimate the equation Y=Beta_0 + Beta_1*X+res
I would like to know if there is a way to impose the estimation of Betas under the constraint of non negative Y (ie: Y>0) in the fitting process (without taking Log of Y). Even looking at the optimize function, I didn't find a way to do it.

2. Assuming that I estimated an equation Y_hat=Beta_0_hat+Beta_1_hat*X
I would like to reparametrize the equation using the following constrained optimization: (see image attached) where Y_hat denotes the dependent variable estimated above and Y_tilde is the closest possible approximation subject to the constraint of positive Y_tilde.
In a nutshell, I want to minimize an objective function subject to an additional constraint.
How can I do this in EViews ?


Many thanks for your answers,

Regards,

Roman

Re: Constrained least square and optimization

Posted: Tue Oct 27, 2015 3:17 pm
by EViews Gareth
EViews does not have constrained estimation or optimization built in. The only way to do it is to perform a transformation such as taking logs.

Re: Constrained least square and optimization

Posted: Tue Nov 26, 2019 10:27 pm
by saranya
is this the case with EViews 11 as well?

Re: Constrained least square and optimization

Posted: Tue Nov 26, 2019 10:41 pm
by EViews Gareth
Yep