Fuzzy Time Series Modeling for Predicting Economic Trends: A Mathematical Exploration

Abstract

In fact, this research investigates the potential use of FTS models for economic indicators prediction in the Asian area, particularly the GDP growth. However, traditional forecasting approaches tend to fail to represent the uncertainty and nonlinearity of the economic data, especially in crisis periods. By applying fuzzy inference rules and defuzzification to create crisp forecasts from fuzzy sets, FTS models capitalize on the principles of fuzzy logic to provide a versatile and adjustable framework for predicting economic outcomes. Using FTS for GDP growth forecasting, the study utilizes a dataset that includes GDP growth and inflation rates over the years 2010 to 2022. FTS models adapt well to uncertain and volatile economic variables, and although challenges in determining the most suitable fuzzy membership function and fuzzy rules exist, the results demonstrate their potential to extract process understanding. The model gave an accuracy (Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE)) of about 5.52 and 5.81, which shows okay predictions, but also we can work on this, and improve our model. To demonstrate the performance of the FTS, we compare to traditional econometric models and machine learning techniques, and highlight the strengths and weaknesses of FTS. This study finally concludes that there is significant potential for evolving FTS models for economic forecasting while future research avenues can assist in developing hybrid fuzzy models, fuzzy clustering and real-time forecasting using FTS models.