Ifs and Soda Based Computational Method for Fuzzy Time Series Forecasting

In the past, many researchers have developed intuitionistic fuzzy set (IFS) and clustering based fuzzy time series (FTS) forecasting methods to enhance the accuracy in forecast in the environment of uncertainty and non-determinacy. The current research work offers a computational approach for FTS forecasting with conjunction of IFS and a self-organized direction aware (SODA) data partitioning algorithm. Proposed FTS forecasting method uses SODA for splitting the universe of discourse (UOD) and IFS based computational method for forecasting. The proposed computational method also uses grey wolf optimizer (GWO) method for optimizing the weights associated with the forecasting parameters because of its unique ability of maintaining a balance between exploration and exploitation processes, reducing local trapping, and speeding up convergence. The usage of SODA aids in making the proposed forecasting method as autonomous as feasible, as it does not require human intervention or prior knowledge of the time series data. Performance of proposed FTS forecasting method is tested in terms of root mean square error (RMSE), average forecasting error (AFE) Mean Absolute deviation (MAD) in forecasting of the enrolments of the University of Alabama. Suitability of proposed FTS forecasting method is also shown by implementing it on a financial time series data of price of state bank of India (SBI) at Bombay stock exchange (BSE), India. Outperformance of the proposed method is also verified with various existing recent FTS forecasting method in terms of RMSE, AFER, MAD, correlation coefficient and Coefficient of determination