Inflation Forecasting for Riau Province: A Comparison of Parametric ARIMA with Nonparametric Nadaraya-Watson Kernel Regression and B-Spline Methods

Forecasting inflation is crucial for assisting in the creation of sound economic strategies, particularly in key regions like Riau Province, one of the country's hubs for the production of crude oil and palm oil. The economic peculiarities of Riau, which set it apart from other regions, lead to inflation patterns in this province that tend to deviate from the national average. Forecasting techniques are crucial for decision-making that promotes stability and overall regional economic planning. The parametric ARIMA method and the nonparametric Nadaraya-Watson Kernel Regression and B-Spline are used in this study to forecast inflation in Riau Province in 2025. While ARIMA is based on certain model assumptions, nonparametric methods are more flexible and can capture more complex patterns. The forecasting results using RMSE, ME, MAE, and MASE showed that the Nadaraya-Watson method performed the best out of the three methods tested. The forecasting results with Nadaraya-Watson Kernel Regression showed a stable decline in inflation, from 0.0464% in March to 0.0191% in August 2025.