摘要
针对影响网络直播带货销量因素彼此之间复杂不可控,将人工神经网络引入网络直播销量预测领域。以皮革服装为例进行实证研究,对神经网络层数、输入层、输出层、隐含层神经元节点进行设计,构建了BP神经网络的网络销量预测模型,并进行了不同隐含层节点数的训练和优化测试。结果表明,隐含层节点数为6,神经网络性能最佳,平均绝对值误差为6.79%,满足网络直播销量预测需求。
Aiming at the complex and uncontrollable factors affecting the sales of webcast,artificial neural networks were introduced into the field of live forecasting.leather clothing was used as an example to conduct empirical research and design the number of neural network layers,input layer,output layer,and hidden layer neuron nodes,the network sales prediction model of a BP neural network was built,and different hidden layer node training and optimization tests were carried out in this paper.The results showed that when the number of hidden layer nodes is 6,the performance of the neural network is the best,and the average absolute error is 6.79%,which meets the demand of network live broadcast sales forecast.
作者
张志新
石文奇
程凡
陈志军
ZHANG Zhixin;SHI Wenqi;CHENG Fan;CHEN Zhijun(ShaanXi Institute of International Trade&Commerce,Xi忆an 712046,China;Silk Road Research Center of Cultural and Creative Industry,Xi忆an 712046,China;Wuhan Textile University,Wuhan 430073,China)
出处
《皮革与化工》
CAS
2021年第3期26-30,共5页
Leather And Chemicals
基金
陕西省教育厅科研计划专项项目(19JK0096)
陕西国际商贸学院科研机构建设项目(2019KZ03)。
