Research On Algorithm Of Positive/Negative Co-Location Pattern Mining In Spatial Data

With the rapid development and wide application of spatial data and database,spatial data mining is becoming more and more important.In geospatial,the subsets of spatial features that are often located together are called spatial(positive)co-location patterns.Although join-based,partial join and joinless algorithms are proposed to solve the problem of co-location pattern mining,but these three algorithms all have the problem of consuming too much time.Negative co-location pattern mining is to find out the spatial feature subset with negative correlation.There is little research on this pattern.The current algorithm must find all the co-location patterns before mining the negative co-location patterns.The mining process is complex and time-consuming.In view of the existing problems of positive and negative co-location pattern mining algorithms,this paper does the following research work:(1)The maximal instance algorithm is proposed in this paper.This algorithm introduces a new concept called maximal instance,proposes a method to generate all row instances of colocation by using the maximal instance,and proves the feasibility and effectiveness of this method.The process of generating row instances and co-location patterns in this algorithm does not need join operation,which can save most of the computing time compared with join-based,partial join and joinless algorithms.(2)An improved algorithm of negative co-location patterns is proposed.By analyzing the relationship between positive and negative co-location patterns,the concept of neighbor relational pair and the non-monotonic property of the participation index of the negative colocation patterns are proposed.This algorithm can mine negative co location patterns without mining all co-location patterns,which solves the complex problem of mining negative colocation patterns.(3)The experimental evaluation of the proposed positive and negative co-location pattern mining algorithms is carried out,using the synthetic data and the real data in the national traffic atlas database of the United States.And the co-location relationship between different types of traffic facilities is found,which proves the feasibility of the algorithm.By analyzing the running time of the algorithm,the validity of the algorithm is proved.The last part of this paper summarizes the research process and results,and points out the shortcomings.