DETECTION FOR IRREGULAR DOMAIN OF MULTI DIMENSIONAL VALUED FUNCTIONAL DATA AND ITS APPLICATION TO SENSING DATA

Abstract

In this paper, we propose a method for identifying an irregular domain of multi dimensional data that are observed continuously and its application to sensing data. We partition the data into three subdomains on each dimension - "irregularly high valued domains", "irregularly low valued domains" and "other domains". Then we detect the subdomains in which the irregularities are observed simultaneously. We demonstrate our method with an actual example using sensing data of environmental radio activity level in Fukushima Prefecture.