Discovery Association Rules in Time Series of Hydrology

Mining association rules in hydrological time series will help to leverage the hydrological data more effectively and extract insightful information from the data, in this paper an improved method of mining quantitative association rules is proposed to implement the hydrological analysis, as well as the optimization of the rules. The improved clustering method is exploited for discretization in the process of mining quantitative association rules. It requires acquiring the initial cluster centers by sampling, then merging the initial clusters iteratively, meanwhile computing the cost in the process of merging. Then rules are generated from the classical Apriori algorithm for mining association rules. Furthermore, with respect to redundancy of the rules, optimized association rules are discovered according to the rule structure and the hydrological analysis requirements. In this way, the semantic integrity between data and the rationality of rules are guaranteed. Finally, the experiment results indicate the feasibility and practicability to analyze the time series in hydrology.