Temporal Association Rules Mining Of The Linkage Relationship Between Secondary Indexes

This paper uses temporal association rules mining methods to study the linkage relationship between the secondary index sequences of the Haixi Consumer Confidence Index.According to different interval division methods of the secondary index sequences,this paper uses qualitative,fuzzy and quantitative association rules mining methods to explore the linkage relationship and the hysteresis effect between the secondary indexes.In terms of mining qualitative association rules,this paper does symbolization processing of index sequence according to the index variance firstly,then finds the linkage relationship and hysteresis effect of the secondary indexes by using the Apriori association rules mining method,and the association rules mining method based on temporal odds ratio.In order to solve the problem caused by hard partitioning of the index sequences in symbolization processing,this paper introduces fuzzy association rules mining method and quantitative association rules mining method.In terms of mining fuzzy association rules,the index sequences are classified using fuzzy soft partition and the membership degree matrix is gotten,then we use the membership function to mine fuzzy association rules.In terms of mining quantitative association rules,this paper directly uses the change ratio of the index sequences to do quantitative association rules mining.In the cases of maximizing specific score,quantitative association rules mining method could find more rules as well as richer changing patterns about antecedent and consequent,and has better effects for mining association rules which include hysteresis items.This paper applies a series of temporal association rules mining algorithms to analyze the secondary indexes of Consumer Confidence Index,to understand the influence of each secondary index changes to the other secondary indexes,and to provide new ideas for studying the internal linkage between the secondary indexes.At last,this paper provides suggestions about the recommendation of Consumer Confidence Index according to the results of temporal association rules mining.Compared with previous studies,this paper discusses and improves the method of economic indicators' variation pattern division.This paper uses a variety of temporal association rules mining algorithms based on different interval division methods.By comparing actual mining effects of various algorithms comprehensively,this paper finds that the quantitative association rules mining method is better for mining the linkage relationship of economic temporal indexes.