We present a new, human consistent approach to the summarization of numeric time series aimed at capturing the very essence of what occurs in the data in terms of an increase, decrease and stable values of data, when relevant changes occur, and how long are periods of particular behavior types exemplified by an increase, decrease and a stable period, all with some intensity. We use natural language descriptions that are derived by using tools of Zadeh’s computing with words and perceptions paradigm, notably tools and techniques involving fuzzy linguistic quantifiers. More specifically, we use the idea of fuzzy logic based linguistic summaries of data(bases) in the sense of Yager, later developed by Kacprzyk and Yager, and Kacprzyk, Yager and Zadrożny. We extend that idea to a dynamic setting, to the summarization of trends in time series characterized by: dynamics of change, duration and variability. For the aggregation of partial trends, which is crucial element in the problem considered, we apply the traditional Zadeh’s fuzzy logic based calculus of linguistically quantified propositions and the Sugeno integral. Results obtained are promising.
|Title of host publication||Forging new frontiers|
|Editors||M. Nikravesh, J. Kacprzyk, L.A. Zadeh|
|Place of Publication||Germany|
|Number of pages||19|
|Publication status||Published - 2007|
|Series||Studies in Fuzziness and Soft Computing|