[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Time series prediction
From: |
Lukas Reichlin |
Subject: |
Re: Time series prediction |
Date: |
Mon, 2 Jan 2012 17:33:09 +0100 |
On 02.01.2012, at 15:46, miro99_ale wrote:
> Hi Juan,
> I really don't think ARMAX is what I need, because the unique stream of data
> in which I'm interested in is the previous data of the time series...
> I know exactly what is an ARMA model (and obviously an ARMAX one); my
> problem is regarding how to obtain a model from the set of data
> (timestamp-ordered) and how to obtain a prediction for the future value...
> So, mathematically, I need the value of x(t) having the values of x(t-1),
> x(t-2), etc...
Looks like you're trying to estimate the current process state using
measurements of its output. Have a look at Moving Horizon Estimation and Kalman
filtering in general.
Lukas