The RATS Software Forum

The program I am running is working very well at the beginning, but if I continue to run the same program many many times(why I need to do this?I just change minor settings in the graph part), later it run into this problem.
It could be solved if I restart Rats.

TomDoan wrote:However, if you run part of a program, you might start at a spot which leaves some variable uninitialized or set to a value from later in the program when it needs the values from the start.

For a TVP model, discount is assigned at the begining, what would be the formula for calculating Prior W and V (given either W or V is chosen, EXACT option is used)?

What if the dependent variable and the independent variable are estimated with error so that there is an errors in variables problem? Is there any method of handling this in DLM?

In a time-varying parameters model?

As is typical, errors in Y are easy since they just turn into part of the measurement error. Errors in "X" are often handled through state-space models since they offer a way to deal with dynamic latent variables. However, if you have both errors in X and in the coefficients on X, you end up with a multiplicative combination of the errors, so it can't be done with a simple DLM. Instead, it needs some form of extended Kalman filter.

TomDoan wrote:In a time-varying parameters model?

As is typical, errors in Y are easy since they just turn into part of the measurement error. Errors in "X" are often handled through state-space models since they offer a way to deal with dynamic latent variables. However, if you have both errors in X and in the coefficients on X, you end up with a multiplicative combination of the errors, so it can't be done with a simple DLM. Instead, it needs some form of extended Kalman filter.

Thanks. That helps me a lot.