Analysis and Modeling of Time Series Based on Granular Computing

Abstract

Abstract-Granular computing is an important theory in solving complicated problems. Currently, researches of granular computing on time series stress more on the design of algorithms, the modeling of time series using information granular is limited. This paper proposed a granular-based model called Temporal Information System (TIS) on time series by using equivalent relationship, which is an extension of granular computing. Major works in modeling include the granulation of time series, development of granular layer as well as the transition rules. The granular operation and its properties provide a novel method for further research on time series which are complicated and uncertain in real applications.