Closed Vera1996 closed 3 years ago
You can do it if you define your model using SSMcustom
components i.e. you define the necessary system matrices manually, and then create corresponding updating function for fitSSM
. Basically, you have to modify Z and T matrices according to your needs so that the previous state (regression coefficient) affects the new state (level and/or slope) via T.
Dear Helske,
is it possible to put the regression effect in the updating equation for the level \mu_t or slope \nu_t? The regression effect is now based on the observations when I use SSMregression. I want to know if de data that I use for the regression, affect the level or slope. So the growth or the growth of the growth. Now I have: model <- SSModel(covid_new[, 1:4] ~ SSMtrend(2, Q = list(diag(NA, 4, 4), diag(NA, 4, 4)), P1 = diag(100, 8, 8), P1inf = diag(0, 8, 8))
I hope you can help me!