Statistica Applicata Vol. 17, n. 2, 2005 159 Giuseppe Giordano 1 Dipartimento di Scienze Economiche e Statistiche, Università di Salerno Via Ponte Don Melillo, 84084, Fisciano (SA) (ITALY) ggiordan@unisa.it Cristina Davino Dipartimento di Studi sullo Sviluppo Economico, Università di Macerata Piazza Oberdan 3, 62100, Macerata (ITALY) cdavino@unimc.it Abstract Aim of this paper is to present a way to explore time series by combining some fundamen- tal results of stochastic processes theory with graphical and reduction features of facto- rial methods. A multiple time series visualization and identification strategy is provided by defining a common structural subspace where different regions related to particular ARMA processes are represented. This subspacebecomes the reference map for the ex- ploration of multiple time series. In order to provide a unique identification of the ARMA model, a complementary tool represented by a classification tree is proposed. Keywords: ARMA process, Principal Component Analysis, Classification Trees, Autocor- relation Functions. 1. INTRODUCTION Nowadays the availability of large data-sets observed at different time points high- lights the need to define methods able to deal with multiple time series. In partic- ular, this requirement is fundamental for national statistical institutes whose aim is to provide frequently and quickly information on huge amount of time series. 1 This paper is supported by MIUR grant 40% “Metodi statistici multivariati e di visualizzazione per l’analisi, la sintesi e la valutazione di indicatori di performance” (Research Unit: University of Salerno - Resp.: Professor Maria Rosaria D’Esposito). VISUALIZING AND EXPLORING MULTIPLE TIME SERIES 1. INTRODUCTION