Essay of a Dynamic Regression by Principal Components Model for Correlated Time Series
María José Del Moral, Mariano J. Valderrama · COMPSTAT · 1998
In this paper we analyze the use of dynamic regression by principal components models for correlated time series forecasting. The choice of an appropriate cutting point on input and output series allows us to study their principal component analysis and the selection of a forecasting model. Two basic issues are discussed on studies with simulated and real data: parsimony and principal components selection in the forecasting model.