From local trend extraction to symbolization of time-series

Article type: Research Article

Authors: Aros, D. Calvelo^a | Chambrin, M.C.^b | Pomorski, D.^a

Affiliations: [a] LAIL – UPRESA 8021, P2 Building, University of Lille 1, 59655 Villeneuve d'Ascq, France | [b] University of Lille 2, Lille, France

Correspondence: [*] Corresponding author. E-mail: denis.pomorski@univ-lille1.fr

Abstract: A methodology is proposed for the extraction of local trends from a time-series. It has been designed to suit the needs of interpretation-oriented visualization and symbolization from raw data. After giving implementation details for efficient computation of local trends, a characteristic analysis span is determined for each time-series. The processing results in a rich visual interpretation and a framework for the local symbolization of a time-series in terms of its value and dynamics.

Keywords: trend, regression, span, symbolization

DOI: 10.3233/IDA-2001-5104

Journal: Intelligent Data Analysis, vol. 5, no. 1, pp. 41-57, 2001