A high-order multi-variable Fuzzy Time Series forecasting algorithm based on fuzzy clustering

摘要: A novel high-order multi-variable algorithm (HMV-FTS) is presented.HMV-FTS outperforms existing algorithms of Fuzzy Time Series (FTS).HMV-FTS reduces computational error compared to the algorithms.Robustness HMV-FTS examined by various examples. High-order for Multi-Variable presented based on fuzzy clustering eliminate some well-known problems with FTS algorithms. can handle only one-variable and one-order FTS. does both tasks simultaneously. cannot incorporate information about future value a variable in forecasting process while can. Defuzzification forecast cluster centers or midpoint intervals use are other controversial These eliminated constructing sets from partition matrices letting each data point contribute defuzzification its membership grade sets. In algorithms, one considered as main which forecasted variables secondary; treats all equally more than be at same time. It shown that suitable system identification, interpolation. This accurate popular tools systems such ANFIS, Type II model ARIMA model.