Fuzzy time series (FTS) methods was first introduced by Song and Chissom (1993, 1994) based on the fuzzy set theory proposed by Zadeh (1965). Over the earlier few years, some methods have been presented based on fuzzy time series to forecast real problems, such as forecasting stock market, temperature prediction, forecasting enrolments, disease diagnosing, etc. Traditionally, time series forecasting problems are being solved using a class of Autoregressive moving average models. Being linear statistical models, they cannot build relationship among the nonlinear variables. Calculating the parameters for multivariables is another issue faced by them. The strong relationship among these variables may result in large errors. Furthermore, a model cannot be estimated correctly if the historical data is less. Therefore, this paper, we propose a new fuzzy forecasting model to overcome the
drawbacks of the traditional forecasting models that aim increasing the forecasting accuracy. In our studies, a hybrid forecasting model based on aggregated FTS and Fourier series analysis. Firstly, we propose weighted models to tackle two issues in fuzzy time series forecasting, namely, recurrence and weighting. Then, the using Fourier series to modify the residuals of the weighted FTS for improving the forecasting performance. By
using the enrolment data at the University of Alabama from 1971s to 1992s as the forecasting target, the empirical results show that the proposed model outperforms one of the conventional FTS models