Prediction of Indonesia Strategic Commodity Prices during the COVID-19 Pandemic based on a Simultaneous Comparison of Kernel and Fourier Series Estimator
M. Fariz Fadillah Mardianto, Sediono Sediono, Idrus Syahzaqi, Siti Amelia Dewi Safitri, Nurul Afifah
Abstract
This study describes a new idea about comparing two new estimations in nonparametric regression for multiresponse cases or simultaneously model based on Fourier series and kernel estimators enabling the prediction of Indonesian strategic commodity prices during the COVID-19 pandemic. Based on the National Strategic Food Price Information Center in Indonesia, there are 10 strategic commodities in the agriculture, livestock, fishery, and horticultural sectors, which have had the biggest endowment to secure food supplies and the formation of inflation figures in Indonesia. These commodities include rice, chicken meat, beef, chicken, egg, onion, garlic, chili, cayenne, cooking oil, and sugar. Using the goodness estimator of criteria in nonparametric regression, such as the smaller Generalized Cross-Validation, the smaller Mean Square Error, and the larger determination coefficient (R2), the result of this study is Fourier series estimator to predict the prices of 10 food commodities, simultaneously. When compared with the kernel estimator, the Fourier series estimator meets the criteria of goodness with a smaller Mean Square Error value of 0.052 and a larger determination coefficient of 99.0472%. The selected estimator has very good performance to predict the prices of 10 food commodities, because the prediction result has very small Mean Absolute Percentage Error equaling 0.0443%. This prediction result can be used for the government to monitor and evaluate price fluctuations for 10 commodities so that the stability of national strategic commodities becomes daily consumption to be maintained, especially during the COVID-19 pandemic.