Mining Competitive Pairs Hidden in Co-location Patterns from Dynamic Spatial Databases

Co-location pattern discovery is an important branch in spatial data mining. A spatial co-location pattern represents the subset of spatial features which are frequently located together in a geographic space. However, maybe some features in a co-location get benefit from the others, maybe they just accidentally share the similar environment, or maybe they competitively live in the same environment. In fact, many interesting knowledge have not been discovered. One of them is competitive pairs. Competitive relationship widely exists in nature and society and worthy to research. In this paper, competitive pairs hidden in co-locations are discovered from dynamic spatial databases. At first, competitive participation index which is the measure to show the competitive strength is calculated. After that, the concept of competitive pair is defined. For improving the course of mining competitive pairs, a series of pruning strategies are given. The methods make it possible to discover both competitive pairs and prevalent co-location patterns efficiently. The extensive experiments evaluate the proposed methods with “real + synthetic” data sets and the results show that competitive pairs are interesting and different from prevalent co-locations.