Mining High Utility Co-location Patterns Using the Maximum Clique and the Subsume Index
Wei Guo Song, Qian Qiao · 2020
Mining high utility co-location patterns (HUCP) is a promising technique in spatial data mining because it treats different features with different levels of importance. Existing HUCP mining (HUCPM) algorithms are based on row instances and table instances, leading to high computational cost. To overcome this problem, an HUCPM algorithm based on the subsume index (HUCPM-SI) is developed using a typical high utility itemset mining model. Maximal cliques are first discovered, enabling the spatial database to be transformed into a clique transaction database. The subsume index is then used to mine HUCPs. Tests conducted on synthetic and real datasets demonstrate the advantages of the HUCPM-SI algorithm.