EViews.com

I am new to state space modeling. Can someone help me with a simple problem that I have seen as an application of the Kalmon filter.

An object starts from a cetain psition. It is traveling in a straight line. The speed has a known expected value but there is some uncertaintly. After one unit of time a new position can be computed from the intial position and the speed. There is some error, however, since the speed is subject to error. Also after the one unit of time has elasped, a measurement of the new location is made. There medasurement is also subject to error.

In theory, the two estimates of the new location, one based on the previous location plus the speed and the other from the measurement, can be combined to provide a more acurate estimate of the new location than either coulld by itself.

Could someone please show me how to set this up using a state space object in Eviews?

I believe I have basically the same question.

I have a value that I am trying to estimate using two separate measurements. The true value won't be known until sometime in the future. How do I combine the two estimates in a state space model in Eviews to get a better approximation of the actual value?

Thanks

These are indeed simple examples and you can easily estimate them in EViews via State Space object. However, before you set up a State Space object in EViews, you should formulate your problem and put it into State Space form. The rest is fairly easy and you can find plenty of explanations/examples in the users guide as well as the forum.

Ok, say I have the true value of something but only through last year. Call the series x. I have two independent estimates for this year, y and z. Can you please provide the simple code to set up the state space model? I am looking for something like:

@state [please fill in]
@signal [please fill in]
@signal [please fill in]

I understand I should be able to find the answer searching the posts and reading the user guide, but I am missing something. Much appreciated.

Does this seem reasonable?

@signal y = sv1 + [var=c(1)]
@state sv1 = c(2) + c(3)*z + [var=c(4)]

since x is unknown at the current point in time, it is not included, although I can use it to estimate the coefficients above... is that correct?

dear eviews administrator,

i've tried to run sspace model in auto specification tab with ARMA (2,0), interest rate of in indonesia as dependent variable (in my workfile: lbi_rate) and independent variable is inflation (in my workfile: inflasi). and i got these:

@signal lbi_rate = sv1

@state sv1 = c(2)*sv1(-1) + c(3)*sv2(-1) + [var = exp(c(1))]
@state sv2 = sv1(-1)

am i wrong? please correct me :)