Global Institute for Research & Education (GIFRE)
Many companies for years accumulate business information, hoping that in the future it will help them with complex analytical research of development tendencies of interesting their processes. By specific example of the semi-structured time series there are considered known fuzzy forecasting models which differ in rules of fuzzification and/or de-fuzzification. This paper presents a new approach to defuzzification of outputs of fuzzy time series on the base of applying the fuzzy set point-estimation method. As compared with some well-known defuzzification rules proposed method improves the statistical quality of semi structured time series forecasting.