An enhanced binary symbolic representation for time series data mining based similarity

Dozens of high level representations of time series have been introduced for data mining in the literature. But the problem of the discretization of the original data into symbolic strings is not been well solved. However, in spite of there are dozens of techniques for producing different variants of the symbolic representation, there still have no excellent method to calculate the distance in the symbolic space to achieve a lower bounding distance. In this paper a novel binary symbolic representation called BSAP is proposed. The representation is unique in which it allows dimensionality reduction and it also grants a lower bound distance measure defined on the symbolic representation. The experiments have been performed on synthetic, as well as real data sequences to evaluate the proposed method.