新冠病毒肺炎疫情影响下基于XGBoost的润滑油销量预测

A XGBoost-Based Lubricant Sales Forecasting Method under the Influence of COVID-19

2020年1月暴发的新冠病毒肺炎疫情对交通运输、旅游等行业造成的冲击给润滑油行业销量带来巨大影响。为实现帮助某大型石油化工企业预测2020年2、3、4月份润滑油销量,本研究基于企业内部交易数据及宏观经济指标使用XGBoost算法完成销量预测。为进一步提高疫情影响下的预测精准度,提出了疫情修正因子方法,使得模型准确率得到了提升。本研究使用的XGBoost结合因子修正的方法为企业遇到突发黑天鹅事件时实现短期内的销量预测提供了一种有效可行的解决思路。

The outbreak of COVID-19 in January 2020 significantly impacted industries such as transportation and tourism,causing a substantial effect on lubricant industry sales.To predict lubricant sales for a major petrochemical company in February,March,and April 2020,this study employed the XGBoost algorithm using internal transaction data and macroeconomic indicators.To enhance accuracy amid the pandemic,we introduced an epidemic correction factor,resulting in improved model precision.The XGBoost method combined with factor correction provides an effective solution for short-term sales predictions during unforeseen events.