Problem with the LOGL function (likelihood estimation)
Posted: Thu Jul 16, 2009 10:41 am
Hello!
Sorry for my bad English, I am French.
I am using Eviews 5 and I have a problem using the logl function.
My problem is I can not solve any likelihood maximisation problem because Eviews always indicates a missing values error when I launch the estimation (the error message is: "Missing values in @logl series at current coefficients at observation X").
I tried to estimate likelihood parameters on very simple datasets (with only 3 values, so I am sure there are no NA's) but it has never worked. The likelihood fonction is quite simple:
I checked the residus and logl values, I also paid attention to the sample range and I specified starting points in the likelihood specifications.
But I could not find a solution.
So can you tell me where the error is?
I put the workfile I am working on in attachment
Thank you for your help!
Sorry for my bad English, I am French.
I am using Eviews 5 and I have a problem using the logl function.
My problem is I can not solve any likelihood maximisation problem because Eviews always indicates a missing values error when I launch the estimation (the error message is: "Missing values in @logl series at current coefficients at observation X").
I tried to estimate likelihood parameters on very simple datasets (with only 3 values, so I am sure there are no NA's) but it has never worked. The likelihood fonction is quite simple:
Code: Select all
@logl logl1
res = d(cac40) - c(1)
var = c(2)
@param c(1) 0.5 c(2) 1
logl1 = log(@dnorm(res/@sqrt(var))) - log(var)/2But I could not find a solution.
So can you tell me where the error is?
I put the workfile I am working on in attachment
Thank you for your help!