Mining Maximal Fuzzy Colocation Patterns

Abstract

Spatial colocation pattern mining is to discover the subsets of spatial objects frequently appearing together in adjacent geographic locations. In the existing research, several algorithms were proposed for excavating maximal prevalent colocation patterns. Furthermore, fuzzy neighborhood relationship(FNR) was employed to evaluate the proximity between spatial instances for improving the accuracy of the mining results. However, the approach for discovering the maximal prevalent colocation patterns based on FNR is not studied yet. This paper defines the maximal fuzzy prevalent colocation pattern (MFPCP). We propose a maximal fuzzy prevalent colocation pattern mining algorithm to generate the MFPCPs instead of all of the prevalent colocation patterns. We conduct experiments on the real datasets to evaluate the performance of the proposed algorithm.