Multiple Time Series Models

Multiple time series is a convenience operation for constructing multiple time series models with a common time interval for use as input to a time series regression.

One of the time series models is identified as the target time series of interest. All of the time series output is produced for the target. The other time series are assumed to be correlated with the target. This operation produces backcasts and forecasts on each time series and computes upper and lower confidence bounds for the identified target series. This operation can be used to forecast a wide variety of events, such as rainfall, sales, and customer satisfaction.

In the example of weather forecasting, the temperature and humidity attributes can be considered as the dependent or correlated time series and rainfall can be identified as the target time series.