In this paper we propose modeling technique, which was applied to multivariate time series data that correspond to different spatial locations (spatial time series). ARIMA model class is considered for each location. Forecasting model for new location is built by spatial "connection" of identified models in observed locations. Spatial "connection" is implemented by spatial averaging of the coefficients of models and by ordinary kriging procedure for means. This modeling technique is illustrated by a substantive example using R system.
This work is licensed under a Creative Commons Attribution 4.0 International License.