Adding "weekly" seasonality in arima R

Question

I've found the internet fairly unhelpful with this issue and thought I'd turn. I have the following arima model:

y = data$produced_diff
regs = data[,c(10,11,12)]
    
ar1ma1x = arima(y, xreg=regs, order=c(1,0,1))

Pretty simple. The data is daily and I'm trying to add a weekly seasonality. I'm currently doing it like this:

ar1ma1xS = arima(y, xreg=regs, order=c(1, 0, 1),
            seasonal = c(0, 0, 7))

I ... think this is right? It runs and gives slightly different output. But I find the documentation fairly opaque.

Answer 1

You may find it easier to use the Arima() or auto.arima() functions from the forecast package, than using the arima() function from the stats package.

Either way, your time series y should be a ts object with frequency set to 7.

Setting seasonal = c(0,0,7) is unlikely to give a good model. Usually, the seasonal orders would be no more than 2.

If you used the auto.arima() function, the orders will be selected automatically:

fit <- auto.arima(y, xreg=regs)

Documentation for auto.arima() is here: https://otexts.com/fpp2/arima-r.html