Forecasting CPI in Malaysia: Comparing linear regression, nonlinear regression, and nonlinear programming methods

This study explores factors influencing the Consumer Price Index (CPI) through an analysis of economic indicators and predictive models. It begins with normality testing and correlation analysis to identify significant variables, followed by model fitting using Linear Regression Model (LRM), Nonlinear Regression Model (NRM), and Nonlinear Programming (NLP). The results show strong positive correlations between CPI and variables like the Coincident Index, Labour, and Volume. Model comparisons indicate that NRM is the most effective predictor of CPI, with slightly lower Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE) values than LRM and NLP. While NLP uses fewer variables, it may simplify model interpretation and reduce computational complexity. This research highlights the importance of accurate predictive models in CPI forecasting for evidence-based policymaking. A limitation is the small dataset, suggesting future studies could explore alternative models, use larger datasets, or conduct simulations to enhance CPI prediction accuracy.