Computer Engineering and Applications ›› 2022, Vol. 58 ›› Issue (24): 276-283.DOI: 10.3778/j.issn.1002-8331.2106-0259

• Engineering and Applications • Previous Articles     Next Articles

E-Commerce Demand Forecasting Based on LSTM Model with Consideration of Hysteresis

BAO Jixiang, LI Lin, ZHAO Mengge   

  1. Business School, University of Shanghai for Science&Technology, Shanghai 200093, China
  • Online:2022-12-15 Published:2022-12-15



  1. 上海理工大学 管理学院,上海 200093

Abstract: The fragmented browsing time and more price-sensitive features of consumer online shopping bring lagging consumption. In order to grasp customer consumption trends, the impact of consumer behavior on sales data is analyzed by obtaining historical sales data and consumer purchase behavior data of paper products for Q enterprise. And random forest is used to select characteristic factors that do not consider hysteresis and consider hysteresis respectively. The demand forecasting model for fast-moving consumer goods is established based on LSTM neural network. According to the data of paper products of Q enterprise, the prediction and verification are carried out. The results show that the relative error of the LSTM model is smaller and the prediction accuracy is the higher when considering the hysteresis.

Key words: long short-term memory(LSTM), hysteresis, e-commerce demand forecasting, random forest

摘要: 消费者网络购物浏览时间碎片化、对价格更敏感的特征带来滞后性消费。为了掌握顾客消费趋势,通过获取Q企业纸类商品的历史销售数据和消费者购买行为数据,分析消费者行为对销售数据的影响,并利用随机森林分别选取不考虑滞后性和考虑滞后性的特征因子;基于LSTM神经网络建立快消品的需求预测模型;根据Q企业纸类商品的数据进行预测及验证,结果表明考虑滞后性LSTM模型预测相对误差更小,预测精度更高。

关键词: 长短期记忆人工神经网络(LSTM), 滞后性, 电商需求预测, 随机森林