Spatial Colocation Pattern Mining with the Maximum Membership Threshold
Meijiao Wang, Lizhen Wang, Zhou Li-hua · Frontiers in artificial intelligence and applications · 2019
Spatial colocation pattern mining discovers the subsets of spatial features frequently observed together in nearby geographic space. The existing colocation pattern mining approaches ignore the proximity level between the instances by a user-specified distance threshold, and return a prevalent colocation pattern set under the single distance threshold. And in many cases, people may want to know the distance range in which a colocation pattern is prevalent instead of a single distance threshold. In this paper, fuzzy neighbor relationship (FNR) is introduced to measure the proximity level between instances by fuzzy set theory. We propose the algorithm for prevalent colocation pattern mining at a single membership threshold (PCP-SMT), and develop the basic algorithm for prevalent colocation pattern mining with the maximum membership threshold (PCP-MMT). Furthermore, an improved strategy is adopted for the PCP-MMT algorithm. The effectiveness and efficiency of the proposed algorithms and the optimization technique are evaluated with an extensive set of experiments.