Development of an Invisible Quality Grade Estimation Algorithm for Facilitating Online Transaction of Heterogeneous Products: The Case of Apple in Agricultural Wholesale Market

This study aims to develop reliable agricultural product grade estimation model based on the wholesale market auction price to facilitate agricultural product e-commerce. Agricultural products are heterogeneous products comparing with commodity products such as, oil and paper. Thus, standardized grade information based on the grade estimation model that this study developed, is helpful to facilitate agricultural product e-commerce. To develop reliable agricultural product grade estimation model, Gaussian Mixture Model (GMM) was adopted as a research methodology. GMM is a useful to analyze a mixed form of several waves or distributions, such as recognition of environmental sounds. Therefore, GMM was selected to develop the grade estimation model in this study. Developed model was empirically tested with the apple auction prices. The results showed that the developed model is consistent with the forecasted grade information by auctioneer as field experts. Specifically, the model showd 82.7% of full coverage rate with the daily data. Thus, agricultural wholesale market system operators and governmental agencies can derive more valuable information from auction price data using the proposed model. It is hoped that future research extend the results of this study to consider various products in an other sectors.