Analysis of Fuzzy Linear Regression and Multiple Linear Regression on Household Income and Expenditure in Malaysia

Abstract

This study examines the relationship between household income and expenditure in Malaysia using Multiple Linear Regression (MLR) and Fuzzy Linear Regression (FLR) models. Household financial data from the 2023 Household Income and Expenditure Survey (HIES) was analysed to assess model performance using the Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). The results indicate a strong positive correlation (0.9117) between household income and expenditure, supporting the hypothesis that increased income leads to higher expenditure. Among the models tested, FLR using Tanaka’s method demonstrated the best predictive performance, with the lowest AIC (2688.11) and BIC (2697.34) values, outperforming both MLR and FLR (Ni) in handling data uncertainty. In contrast, MLR exhibited significantly higher AIC (120064042) and BIC (120064048.2) values, suggesting a poorer fit for uncertain economic data. These findings highlight the advantages of FLR in modelling complex economic relationships and suggest its potential applications in economic forecasting. Future research should focus on refining FLR techniques further to enhance their accuracy and applicability in economic analysis.