Research of spatial co-location pattern mining based on segmentation threshold weight for big dataset
Minwei Tang, Zhanquan Wang · 2015
Discovering spatial co-location patterns is an important field in many scenarios such as identifying plant distribution, detecting crime, etc. In practical applications, all attributes on a geographic surface are related to each other, but closer values are more strongly related than the distant ones. The distance scale between two instances of different types and the big dataset aren't considered in the previous co-location pattern mining algorithms. A new co-location pattern mining approach is proposed based on segmentation threshold weight which is meant to measure the relation of two instances according to their Euclidean distance. This approach defines a new weight interest measurement. The new pruning technique is used to reduce the amount of candidates and the algorithm is realized in the MapReduce framework and run on the distributed platform Hadoop according to the proposed guiding strategy. The correctness, completeness and efficiency of the proposed methods are analysed and the experiments in real and synthetic dataset have been performed to prove that the algorithm is effective and feasible.