Data Mining on Nonlinear Temporal Data
Weitong Huang, Mingyu Lu, Zixiang Zhao · 2006
One of the goals of data mining is to discover hidden rules from existing data. Real rules in data differ according to characteristics of the data, and the effect of data mining depends mostly on whether the method selected matches the characteristics of the data. To improve effect of data mining, this paper discusses first correlation of data mining methods and characteristics of data taking temporal data generated from dynamics system as an example, then types of dynamics system since characteristics of data are determined by type of dynamics system and how to determine them from the data. At last we build a neural network to mine the data given the type and parameters of the dynamics system