基于FTRL和XGBoost组合算法的电商销量预测系统

随着近几年电商行业的大规模发展,越来越多的商品正通过各电商平台销售,成交量越来越巨大。在2019年"双十一"期间,仅天猫一家平台的成交额就达到2684亿元人民币。为了应对急剧增加的成交量,就需要制定合理的库存计划,最大限度避免库存积压、尾单、缺货等现象,减少企业的缺货成本和库存成本,从而提高企业利润。为了制定合理的库存计划,提前对销量做出准确的预测就显得尤为重要。本文提出了一套基于FTRL+XGBoost算法的电商销量预测系统,与LSTM算法进行了销量预测准确率和实时性的对比,结果验证了系统的有效性。

E-commerce sales forecasting system based on FTRL and XGBoost algorithm

With the development of the e-commerce industry in recent years,more and more commodities are being sold through various e-commerce platforms,with more and more transactions.During the singles’ day event in 2019,the transaction volume of Tmall reached 268.4 billion yuan.In order to cope with the sharp increase in orders,it is necessary to develop a reasonable inventory plan.It can avoid the phenomenon of overstock,final order,stock shortage and so on,so as to improve the enterprise’s profit.In order to make a reasonable inventory plan,it is particularly important to make an accurate forecast of sales in advance.From the perspective of the system,this paper proposes a set of e-commerce sales forecasting system,which improves the accuracy and timeliness of sales forecasting,and the effectiveness of the system is verified by comparing the results of various algorithms.