关键词: 时间序列相似性, 包络线距离法, 模糊集合

Abstract: This paper presents a new method of fuzzy envelop distance for comparing time series similarity to improve the widely used envelop distance with fuzzifying two clear discriminant boundaries, thus not only reducing the sensitivity of the discriminant results to boundary parameter values, but also making the discriminant method more robust. The new method is then applied to recognize the economic meaning of the macroeconomic independent factors by comparing the similarity between the independent components and economic indicators. And the results, compared with that of the original envelop distance, show that the performance of similarity mining between time series is highly improved.

Key words: time series similarity, envelop distance, fuzzy sets