# How to estimate a multivariate model with marima in R?

I am going through the examples on how to use the marima package in R provided here.

"We will estimate the four univariate the models [...] In order to define the model theprocedure define.model is used [...] Using define.model the variables in the data which are irrelevant for the analyses are taken out, rem.var=c(1,6,7), and indep=c(2:5) results in the variables 2, 3, 4 and 5 being analysed independently. "

Later on:"We may now estimate the general four-variate arma(1,1) model. The only modification in comparison with the above analysis is the model (Model2) definition statement where indep=c(2:5) is taken out (indep=NULL)."

When I analyse the variables independently using arima am I evaluating a multivariate model? Or in order to evaluate a multivariate model I need the condition indep=NULL?

• The standard term for multivariate ARIMA is vector ARIMA, i.e. VARIMA, just like the standard term for a multivariate autoregression is vector autoregression, i.e. VAR. This may help you find relevant information easier. – Richard Hardy Dec 8 '17 at 17:59