Time Series Classification with Linguistic Summaries

Abstract

Soft computing techniques may provide various forms of human-consistent summaries about large time series databases, e.g., linguistic summaries, frequent patterns, fuzzy IF-THEN rules. Within this research, we focus on linguistic summaries constructed as linguistically quantified propositions, that may be exemplified by ‘Among all increasing trends, most are short’. We pose the question whether such imprecise results of summarization may successfully support the classification of time series data. Within the proposed approach, we classify a vector of linguistic summaries instead of classifying crisp time series. The approach is illustrated with experiments on artificial and benchmark real-life time series datasets. It turns out to be very promising for the classification of autoregressive time series by the probabilistic models.