The Program for Fuzzy Time Series Forecasting Based on Z-numbers

Forecasting means predicting values for the future based on previous regularities. The formation of forecasts is based on the statistically emerging trends in the change of certain numerical values ​​of the object. For the most part, any information people use for making decisions is presented with vagueness and a degree of reliability. Usually people are not certain in information they provide. Nevertheless, in the most popular and widespread methods for forecasting using fuzzy logic, data is used AS-IS. Based on the idea that information has different level of confidence degree of reliability, Lotfi Zadeh introduced a new concept called Z-numbers. Z-numbers include the perception of uncertainty of the information and consists of two components: a constraint on value and a degree of certainty. The proposed approach uses the modification of Z-numbers based on perception. It is shown, where and why Z-numbers can appear and how to work with them. As a result, the idea of ​​using Z-numbers based on perception at the step of fuzzification is suggested, as well as how to process this representation of fuzzy numbers based on perception. An algorithm for predicting time series using Z-numbers is proposed. The paper contains 55 pages, 3 chapters, 61 pictures, 15 items in bibliography list, 4 sections in appendix. Key Words: forecasting, fuzzy time series, fuzzy logic, Z-numbers