A New Insight into the Linguistic Summarization of Time Series Via a Degree of Support - Elimination of Infrequent Patterns

We extend our previous works on using a fuzzy logic based calculus of linguistically quantified propositions for linguistic summarization of time series (cf. Kacprzyk, wilbik and zadrożny [4, 5, 6, 7, 8, 9, 10, 11, 12, 13]. That approach, using the classic degree of truth (validity) to be maximized, is here extended by adding a degree of support. On the one hand, this can reflect in natural language the essence of traditional statistical approaches, and on the other hand, can help discard linguistic summaries with a high degree of truth but a low degree of support so that they concern infrequently occurring patterns and may be uninteresting. We show an application to the absolute performance analysis of an investment (mutual) fund.keywordstime seriesfuzzy logiconline algorithmmean absolute deviationmining time series datathese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.