Finding Persisting States for Knowledge Discovery in Time Series

Abstract

Knowledge Discovery in time series usually requires symbolic time se-ries. Many discretization methods that convert numeric time series to symbolic time series ignore the temporal order of values. This often leads to symbols that do not correspond to states of the process generating the time series. We propose a new method for meaningful unsupervised discretization of numeric time series called " Persist " , based on the Kullback-Leibler divergence between the marginal and the self-transition probability distributions of the discretization symbols. In evaluations with artificial and real life data it clearly outperforms existing methods.