The Application Of Association Rule In The Data Mining On Time Series

Time series is a set whose data has been arranged in time. The time series analysis is a process which get the useful conclusion by analyze the time series. there are a lot of methods, for example, Auto-Regression Model(AR),Moving Average Model(MA),Auto-Regression Moving Average Model and Data Mining,Higher-Order Statistics which has been developed in recent years.Data mining, which has been considered as a important methods in the analysis of time series, received more attention came from boffin. Data mining is a process which get the useful information from the vast,incomplete,noised,fuzzy and random data.Association rules, which was considered as one of the most important methods of data mining, has been applied to data mining on time series. Its aims at discovering the interest association of items on data. Apriori algorithms, which was advanced by Rakesh Agrawal, is the most classic algorithms.It has only one fixed support and confidence in Apriori algorithms. there are two problems in the concrete data mining, one is the support, that is to say, we will not cover enough data to be used in finding the intrest association rules if we set a high support, but we will find a lot of useless association rules if we set a low support. The other is confidence, that is, we take a fixed confidence during data mining based on the similar reasoning intensity of every rules, but the condition is not a realism in practice.This paper put forward the improved algorithms named Mean Threshold Apriori(MT-Apriori) in the mining of association rules on the basis of domestic and international research on association rules, it takes a mean support in the mining of frequent items and takes mean confidence in the mining of association rules, we can avoid the problems and mine the interest association rules in this algorithms. In the end of this paper, this algorithms was applied to stocks analysis,supermarket data and hospital data, it has a better performance in the mining of association rules than the Apriori algorithms.