基于LSTM 神经网络的电商商品销售预测方法  

Sale Forecast Method of E-commerce Products Based on LSTM Neural Network

作　　者：柯苗[1] 黄华国[1] KE Miao;HUANG Huaguo(Fujian Vocational and Technical College of Water Conservancy and Electric Power,Yong'an,Fujian 366000,China)

机构地区：[1]福建水利电力职业技术学院信息工程系,福建永安366000

出　　处：《福建师大福清分校学报》2020年第5期83-89,共7页Journal of Fuqing Branch of Fujian Normal University

基　　金：2019年福建省中青年教师教育科研项目(编号JAT191234).

摘　　要：探讨深度学习技术在时间序列预测方面的理论依据.将商品销量预测归纳为多变量时间序列的预测问题,使用某电商网店的历史销售数据,详细介绍在TensorFlow框架下搭建LSTM网络模型的方法,对比了使用AR模型进行预测的运算结果,得到LSTM网络模型具有数据输入简单方便,在网络的结构、训练方法的效率及有效性和预测的准确性等方面都具有更大的优越性.研究结果对电商企业改善营销决策和合理的库存管理具有重要的指导意义.This paper discusses the theoretical basis of deep learning technology in time series prediction and classifies the prediction of commodity sales into the prediction of multivariate time series.Based on the historical sales data from certain e-commerce shop,this paper has introduced the method of building LSTM network model in TensorFlow framework.By comparing with the result from the prediction of AR model,it shows that LSTM network model has the advantages of simple and convenient data input,network structure,training method efficiency and effectiveness,prediction accuracy and so on.The results of this paper have important guiding significance for e-commerce enterprises on improving marketing decision-making and reasonable inventory management.

关 键 词：商品销量预测 LSTM 网络模型搭建 

分 类 号：F713.3[经济管理—产业经济]