Clark West MSPE adjusted statistics
Posted: Sun Mar 03, 2013 5:37 am
Hello,
I changed the question a bit, Ithink nobody knows much about this.
I am trying to measure the performance of a few forecasting models I have.
There are 2 adjusted statistics: DM and CW. I checked several sources but I am not sure about the difference. I wonder if anybody has an idea to calcylate the Clark and West adjusted statistics.
From what I read here is Diebold-Mariano statistics: S= d/sqrt(varience of d(t)) sqrt(N)
d(t) = RMSE1(t) - RMSE2(t)
mean of d(t) ==> d= 1/N sum of (d(t)
Thanks a lot
I changed the question a bit, Ithink nobody knows much about this.
I am trying to measure the performance of a few forecasting models I have.
There are 2 adjusted statistics: DM and CW. I checked several sources but I am not sure about the difference. I wonder if anybody has an idea to calcylate the Clark and West adjusted statistics.
From what I read here is Diebold-Mariano statistics: S= d/sqrt(varience of d(t)) sqrt(N)
d(t) = RMSE1(t) - RMSE2(t)
mean of d(t) ==> d= 1/N sum of (d(t)
Thanks a lot