摘要
针对新零售行业为了满足市场需求,逐步转入多品种小批量的新生产模式,带来前所未有的库存管理难度,提出了一种基于数据分析实现对销售量预测的方法。该方法采用Pearson相关系数分析变量之间的相关性,对相关性进行假设检验。得出显著相关因素,并建立BP神经网络预测模型预测销售量,从Matlab平台上得到的仿真和测试结果表明,该模型的预测值误差MAPE小于0.006,因此该方法可以帮助新零售公司更好地掌握市场需求,安排库存,从而降低销售成本。
In order to meet the market demand,the new retail industry gradually turn into many new varieties of small batch production mode,brings unprecedented difficulty in inventory management,a forecasting method based on data analysis for sales is proposed.In this method,Pearson correlation coefficient is used to analyze the correlation between variables,and finally the correlation is tested by hypothesis.Significant correlation factors were obtained and a BP neural network prediction model was established to predict sales volume.Simulation and test results obtained from Matlab platform showed that the predicted value error MAPE of the calculated model was less than 0.006.Therefore,this method can help new retail companies meet the market demand better,arrange the inventory,and thus reduce the selling cost.
作者
肖泉彬
黎小龙
车俊俊
何敏
XIAO Quanbin;LI Xiaolong;CHE Junjun;HE Min(School of Software Engineering,Jiangxi University of Science and Technology,Nanchang 330013,China)
出处
《电子设计工程》
2021年第17期109-111,116,共4页
Electronic Design Engineering
关键词
大数据挖掘
相关性分析
BP神经网络
因子修正
big data mining
correlation analysis
BP neural network
factor correction
