Page 1 of 1

F Stat Probability Value / data member

Posted: Sun Mar 05, 2017 11:20 pm
by diggetybo
I was looking at the documentation of the equation object's data members on pg 35 of the object reference. It seemed to have everything I would ever need. However I was trying to find a data member for the f stat's probability, and I was unable to find such a data member. There was at least @f for the f-stat itself, but I'm trying to examine joint significance over time.

Specifically, I'm doing a rolling regression and after each iteration I will need to store the f-stat's probability. I suppose I could work it out arithmetically, but my data set is large and I want to save on computational resources as much as possible. Calculating the f-stat's probability from scratch at each step of the rolling regression seems intensive. Although I'm actually not sure how severe the trade off is in performance when using built-in data members versus computing everything from scratch. If it's negligible maybe I'll just bite the bullet, but if there is a better way please let me know.

Thank you

Re: F Stat Probability Value / data member

Posted: Mon Mar 06, 2017 9:32 am
by EViews Gareth

FWIW, the computational resource would have been negligible. I just wrote an EViews program that calculated 1,000,000 f-probs in .6 of a second.

Re: F Stat Probability Value / data member

Posted: Mon Mar 06, 2017 10:15 am
by diggetybo
Thank you for looking into all aspects of my question, I've always wondered about the efficiency aspect. Now I know I shouldn't worry so much. :)