An Imprecise High-Order Fuzzy Time Series Model Forecasting the Stocks Traded Using Central-Log-Ratio Transformation of Compositional Data

Malaysia, Indonesia, and Thailand countries have half of the trade in the group of the Association of Southeast Asian Nations (ASEAN) members. An agreement has increased the importance of trade coalition without the US dollar among them. To understand the pattern of the Stock Traded in these countries, this study developed ample fuzzy time series foretelling models. An innovative data transformation approach called compositional data is employed. The Fuzzy Time series models are implemented with different orders of fuzzy logical relationships. The Trapezoidal and Spline S-shaped membership functions are engaged in these models. The performance of forecasting is evaluated through the compositional root mean square error (CRMSE) and compositional mean absolute percentage error (CMAPE). The analysis of forecasted accuracy measurements showed that the Third-Order Fuzzy Time Series model with a Trapezoidal membership function outclassed other orders models. It is also observable that Thailand's stock traded values increased compared to Malaysia and Indonesia.