Nurhayadi, Subanar, Abdurakhman, Agus Maman Abadi
Many scientists have been studied time series modeling using fuzzy method. Fuzzy method has an advantage in time series modeling because it permits linguistic variable as an input, namely economists' experiences. This paper discusses model transformation in the extent of median error. Sliding the model in the extent of median error directing to zero is proven to be able to minimize Mean Absolute Error (MAE). Comparison with other methods to predict the enrollment of new students at the University of Alabama, has shown that our proposed method can provide the smallest MAE. In the case study of stock price prediction of Bank Mandiri (Persero) Tbk (BMRI.JK), the translation of weighted Takagi Sugeno Kang fuzzy model of zero order can reduce MAE. © 2014 Nurhayadi et al.
Department of Mathematics Education, Tadulako University, Indonesia; Department of Mathematics, Gadjah Mada University, Indonesia; Department of Mathematics Education, Yogyakarta State University, Indonesia
