Extracting Spatial High Utility Co-location Patterns Based on Fuzzy Feature Clusters

Sets of spatial features whose instances frequently appear together in nearby areas are regarded as spatial co-location patterns (SCPs). Spatial high utility co-location pattern(SHUCP), which is an extended research of SCPs, can better reflect the high quality aggregation phenomenon among spatial features. Utility participation index (UPI) is a commonly used adaptive utility measurement in most of the existing methods. Unfortunately, UPI does not take into account the impacts of fuzziness and overlap of proximity relationships on the calculation of utility. Additionally, UPI mainly relies on the computation of the instance layer and does not satisfy the downward closure property. In this paper, we combines fuzzy set theory to establish fuzzy neighbor relationships and discuss the issue of instance overlap between neighbor relationships. Then, a new calculation method for the utility index between spatial features has been design with considering fuzzy contribution score of instances. Using Fuzzy Chameleon Clustering algorithm (FCC) based on fuzzy utility index, this paper further extracts SHUCPs from different high utility fuzzy feature clusters. Experiments are conducted on three synthetic data sets and two real data sets to prove the rationality and effectiveness of the proposed algorithm.