Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.
|
|
|
|
Re: statistical testing of optimized parameter
Posted:
May 20, 2012 3:11 PM
|
|
"Bruno Luong" <b.luong@fogale.findmycountry> wrote in message <jp5l69$jic$1@newscl01ah.mathworks.com>...
>
> > I don't understand how it could help me to find parameter covariance matrix. Can you please elaborate a little?
>
> Assuming we have the linear relation
>
> p = f(d) = A*d
>
> d is the data, A is the inversion mode, p is the parameters. You ask what is the covariance if p, then you need to know the covariance of d. I don't see anywhere you have it in the code.
>
> So it is quite normal what you compute is wrong.
>
> > I am sorry for asking that but I am not good in all those mathematical theories. So, it would be really helpful if you can explain a bit more or suggest me any simple resources which can help me to sort out.
>
> You don't need yet to look anywhere. Foremost you have to ask the right question. If you don't know the noise of your data, there is no point trying to compute the covariance of the parameters.
>
> Bruno
Hi Bruno,
Sorry for this late message. I understand what you mean!
I tried to find out the covariance of my data. But the problem is my main model (velocity) has four dependent variables (I showed it in my previous mail) and among them only V(1-10) is using real data but rest of them are simulated data. You see, my model is a sort of ARMAX type model where the first term of the model is the model output of previous time step. As far as I understand, I will need a matrix form of my explanatory variables but in this case I don't know how to do that. Can you help me?
- Hasan
|
|
|
|
