KNFCOM-T: a k-nearest features-based co-location pattern mining algorithm for large spatial data sets by using T-trees

Abstract

Spatial co-location patterns represent the subsets of Boolean spatial features whose instances often locate in close geographic proximity. The existing co-location pattern mining algorithms aim to find spatial relations based on the distance threshold. However, it is hard to decide the distance threshold for a spatial data set without any prior knowledge. Moreover, spatial data sets are usually not evenly distributed and a single distance value cannot fit an irregularly distributed spatial data set well. In this paper, we propose the notion of the k-nearest features (simply k-NF)-based co-location pattern. The k-NF set of a spatial feature's instances is used to evaluate the spatial relationship between this feature and any other feature. A k-NF-based co-location pattern mining algorithm by using T-tree (KNFCOM-T in short) is further presented to identify the co-location patterns in large spatial data sets. The experimental results show that the KNFCOM-T algorithm is effective and efficient and its complexity is O(n).