Evaluating multiple time series models for consumer price index forecasting to support national defense decision-making

Price stability, as reflected in the Consumer Price Index (CPI), plays a crucial role in supporting economic resilience and national defense readiness. This study evaluates multiple time series forecasting models, including Error-Trend-Seasonal (ETS), Holt, Holt–Winter, SARIMA, SARIMAX with exogenous variables, and hybrid approaches combining Holt/Holt–Winter with SARIMA, to identify the most accurate method for predicting Indonesia’s CPI. Monthly data from 2017–2022 were analyzed using a training–testing split, and forecasting accuracy was assessed based on RMSE. The results show that the Holt–Winter model outperforms all other approaches, achieving the lowest RMSE value of 1.9159. Residual diagnostics confirm that the Holt–Winter model effectively captures trend and seasonal patterns, with errors behaving close to white noise. These findings highlight the superiority of Holt–Winter in providing reliable CPI forecasts, offering significant implications for economic policy formulation and strategic planning in the context of national resilience.