부산대학교 도서관

Agricultural product prices have a great influence on the production value of a country, with significant economic impacts on farmers and consumers. This research uses weather data, international oil price data and social news to conduct sentiment analysis to predict future agricultural product price trends. The resulting data is then displayed using rolling and recursive window methods for segmentation and evaluation. The research results show that adding emotional scores and oil prices to predict agricultural product prices can effectively improve the prediction results. In terms of segmentation performance, the linear regression provides better prediction results than the quantile regression, and the recursive window method provides better prediction results than the rolling window.