In order to ensure the freshness of agricultural products and reduce the cost of loss due to product decay, weather factors such as weather conditions, wind level and air quality index are incorporated into the fresh agricultural product sales forecasting model. Then, based on the historical sales data of agricultural products, three machine learning methods of Ridge Regression, Random Forest and Support Vector Machine are used to perform regression prediction. The prediction results show that the fresh agricultural product sales forecasting model considering weather factors can significantly improve the prediction accuracy. The relative reduction rate of the Root Mean Square Error achieved by the three algorithms is 68.90%, 23.66% and 59.52%. And relative reduction rate of the Mean Absolute Percentage Error is 66.2%, 34.99% and 61.13%, respectively.
Wang, Xuping; Lin, Dongping; Fan, Wenping; and Wang, Tianteng, "Research on Sales Forecast of Fresh Produce Considering Weather Factors" (2018). ICEB 2018 Proceedings (Guilin, China). 47.