New Regional Co-location Pattern Mining Method Using Fuzzy Definition of Neighborhood

Abstract

Regional co-location patterns represent subsets of object types that are located together in space (i.e. region). Discovering regional spatial co-location patterns is an important problem with many application domains. There are different methods in this field but they encounter a big problem: finding a unique optimum neighborhood radius or finding an optimum k value for nearest neighbor features. Here, we developed a method that considers a neighborhood interval using fuzzy definition of neighborhood. It is easier to apply the proposed method for different applications. Also, this method mine regional patterns using a local tessellation (Voronoi Diagram) and finds patterns with a core feature. To test our method we used a synthetic data set and compared developed method with a naïve approach. The results show that the proposed method is more applicable and efficient.