Food price fluctuations occur almost daily and directly affect purchasing power as well as the stability of regional and national economies. As one of the largest provinces in Indonesia, East Java, which significantly contributes to national GDP, has diverse economic structures and highly sensitive to price changes. Given this situation, government needs more accurate prediction methods to monitor Consumer Price Index (CPI) movement as a basis for establishing more appropriate economic strategy and policy. This study aims to compare the performance of Polynomial Regression (PR) and Support Vector Regression (SVR) in predicting CPI using food price data from SISKAPERBAPO for the 2014 - 2020 period, covering regencies and cities in East Java. To ensure the quality of the analysis, missing values were removed. A Pearson’s r correlation analysis was then conducted to assess the relationships between food prices and CPI. The model obtained was then evaluated using mean squared error (MSE), root mean square error (RMSE), Mean absolute percentage error (MAPE), and computation time. The results shows that third order PR achieved higher accuracy with MAPE of 0.3% (training) and 3.4% (testing), while SVR performed lower with MAPE of 5.9% (training) and 6.0% (testing). In addition, PR was more computationally efficient than SVR. These findings underscore PR as a more reliable method for predicting CPI using complex regional food data.
Ayu Adelina Suyono (Sun,) studied this question.