Tomi Yahya Christyawan, M. Syauqi Haris, Rafiuddin Rody, Wayan Firdaus Mahmudy
Fuzzy Time Series is a fuzzy logic implementation method for forecasting based on previous time series data that many researchers have made much effort to improve it. Determining the interval length at the fuzzification stage in Fuzzy Time Series method is very influential on the forecasting result. In this study, Genetic algorithm is applied in the fuzzification phase of the fuzzy time series. This method intends to improve the accuracy of forecasting methods by determining the appropriate objective intervals. A better forecasting result is expected from the dynamic intervals obtained by genetic algorithm method, instead of forecasting result obtained by static intervals. From the optimization test results, fuzzy series time interval obtained by using genetic algorithm is able to produce average of RMSE value as low as 116 from 10 experiments and outperform earlier methods that only have the lowest RMSE at 149.2. © 2018 IEEE.
Faculty of Computer Science, Brawijaya University, Malang, Indonesia
