Research And Application On Quantitative Spatial Association Rule Mining

Spatial association rule mining, an important part of spatial data mining and a branch of the association rules mining, can provide implicit valuable knowledge in the spatial data for people and find spatial dependencies, the interaction and symbiotic relationships. In existential study about spatial association rule mining, many scholars pay more attention to boolean associaion rule mining’s data and algorithms but little to quantitative association rule mining. On the basis of the scholars’previous work, this paper studied the quantitative spatial association rule mining, based on the spatial data preprocessing, and the association rules algorithm to improve the traditional association rules, which are part of the quantitative information, finally, using the simulated annealing algorithm to extract quantitative information on association rules.This paper contains the following contents:Firstly, this paper analyzes research the nowadays status of the spatial association rule mining and the quantitative association rule mining and points out the main problems existing in current research.Secondly, this paper studies the basic concepts and the main algorithm in spatial association rule mining, data preprocessing method and simulated annealing algorithm. Also combaining the multilevel associations rule mining, this paper gave the basic process of quantitative spatial association rule mining.Thirdly, after analyzing the problems in data preprocessing of spatial association rule mining, this paper preprocesses the data by using spatial clustering. The spatial data which is using for spatial association rule mining is discretized by spatial clustering, each cluster has been discretized as a discrete space. So it can retain the implicit spatial information as much as possible when preprocessing the spatial data.Fourthly, FP-Growth and FP-tree are improved for quantitative spatial association rule mining. This paper reconstructs the FP-tree data structure, which contains the transaction information, and the FP-Growth algorithm is improved, so that association rule contains transaction information.Finally, after studying the basic ideas and the process of simulated annealing algorithm, it is taken to mine quantitative association rules from association rules containing the transaction information. A practical example is given to illustrate the rationality of the proposed method.