Two seasonal periods in ARIMA using R

Question

I'm currently using R to predict a time series with these instructions:

X <- ts(datas, frequency=24)
X.arima <- Arima(X, order=c(2,1,0), seasonal=c(1,1,1))
pred <- predict(X.arima, n.ahead=24)
plot.ts(pred$pred)

As you can see I've data each hour, and I chose the seasonal period of 24 (one day).

I would like to improve my forecasting using an additional seasonal period in order to include the seasonal component of the week (seasonal length of 7*24=168 data)

Is there any method for this? How do you do it?

UPDATE: I've read this (your) blog page, maybe can I use the external regressors to simulate a second seasonal period?

Answer 1

There are no R packages that handle multiple seasonality for ARIMA models as far as I know. You could try the forecast package which implements multiple seasonality using models based on exponential smoothing. The dshw, bats and tbats functions will all handle data with two seasonal periods.

Answer 2

I found this paper:

• Au, et al. Automatic Forecasting of Double Seasonal Time Series with Applications on Mobility Network Traffic Prediction

It is about predicting the mobile network traffic prediction using the automatic double seasonal ARIMA. As it is a research paper, it has clearly described the algorithm that one can adopt to adopt multi-seasonal ARIMA prediction. So far, it has given me enough background to proceed further with my research.