LASSO를 이용한 냉동 오징어 소매가격 예측 모형 추정

Abstract

This study predicted the retail price of frozen squid using LASSO. In addition, it was predicted using the Univariate model ARIMA and the Multivariate model ARIMA-X, and VAR. Thereafter, predictive power between models was compared based on the results of the estimated models. In the case of the LASSO, it was divided into two types: static forcast and dynamic forcast. As a result of the prediction, it was found that the results of static forecast were superior to those of dynamic forecast. Among the models which performed static forecast, the prediction results of Adaptive LASSO to predict 1 month later were the best. In the univariate models, the ARIMA(2,1,2) model had the best predictive power, and the ARIMA-X(2,1,2) made the most accurate prediction in the multivariate models. Based on the prediction results, the MDM test showed that the predictive power of Adaptive LASSO with static forecast was superior to that of LASSO with dynamic forecast, ARIMA(2,1,2), Multivariate model ARIMA-X(2,1,2), and VAR(5).