New Algorithm for Association Rule Mining in Trend Fluctuation Based on Concept Lattice
Yong Zhang · 2008
This paper presented a method of cycle association rules mining for time-series fluctuations based on the concept lattice theory.The time-series were first deseasonalized and then a new parallel algorithm of cycle association rules mining was proposed.To increase the mining speed efficiently,some concepts obtained in the algorithm were pruned.Afterward,deseasonalizeation of time-series which do not satisfy the pretreatment conditions of deseasonalizeation in moving average method was computed by a high accuracy model given in this paper.The experiment results show the validity of this method.