Research And Application Of Spatio-temporal Clustering And Association Algorithms For Multi-attribute Values

Spatiotemporal data mining is one of the frontier and emerging research fields of data mining,which aims to analyze spatiotemporal data of higher dimensions and extract potential and valuable knowledge from the spatiotemporal data.Spatiotemporal clustering and spatiotemporal association are two important branches of Spatiotemporal data mining.It is very difficult to set the threshold of spatio-temporal clustering reasonably and apply the spatio-temporal clustering and association rules to the spatio-temporal data sets with multi-attribute dimensions.In view of these problems,the following parts are mainly studied in this paper:Firstly,in order to solve the problem that the traditional spatiotemporal clustering algorithm ST-DBSCAN has a large randomness in artificial threshold setting,which results in unsatisfactory clustering results,a method of setting threshold by spatiotemporal distance-frequency histogram was proposed in this paper,and the clustering results under this method were proved to be more reasonable and accurate through comparative experiments.Secondly,the clustering analysis is carried out in this paper.The ST-DBSCAN of the spatio-temporal clustering algorithm is limited to the processing of three-dimensional spatio-temporal data.For clustering more than three-dimensional spatio-temporal data,a new hybrid attribute spatio-temporal clustering algorithm is proposed.The algorithm calculates the similarity of attribute characteristics between multiple transaction objects.Multiple transaction objects that meet the similarity threshold can be included in the same spatial and temporal cluster by introducing Gower similarity coefficient,Dice similarity coefficient and Euclidean distance to build the mixed attribute similarity model,so the original spatiotemporal clustering algorithm is extended to cluster analysis of more dimensions.Finally,this paper expands the space,time and attribute semantics on the basis of the traditional association rule concept on how to judge the association relationship between the space-time and attributes of the multidimensional spatiotemporal data set on the basis of the association rule algorithm fp-growth,and an improved multi-attribute spatio-temporal association algorithm is proposed,so as to calculatethe association rules between spatio-temporal and attributes of multi-dimensional spatio-temporal data sets considering the constraints of space and time.A multi-dimensional spatio-temporal data sample database was built,and the model was built and the algorithm was implemented to mine the sample database by downloading job recruitment data.The results show that the two algorithms have good effect on clustering and association analysis of multi-attribute spatio-temporal data,and have good universality.If applied to different types of data,the result analysis can provide reasonable,effective and practical guidance.