Generalized shifted-factor analysis method for multivariate geo-referenceddata

Multivariate data with spatial dependencies arise in many areas of applicat ion, including geology, precision agriculture, and ecology. For analysis of such data, a methodology based on a generalized shifted-factor model is de veloped. The model incorporates potential lagged dependencies between facto rs and observed variables, representing asymmetric spatial dependencies obs erved in practice. Identification and estimation issues are discussed. A pr ediction procedure that exploits both the multivariate and spatial dependen ce in the data is proposed and illustrated.