基于SVM的短生命周期产品供应链双渠道需求预测模型  

SVM-based Dual-channel Demand Forecast Model of Perishables in The Supply Chain

作　　者：徐琪[1] 刘峥[1] 

机构地区：[1]东华大学管理学院,上海200051

出　　处：《系统管理学报》2014年第2期255-262,270,共9页Journal of Systems & Management

基　　金：国家自然科学基金资助项目(71172174;70772073);国家公派联合培养博士生项目(留金发[2012]3013);教育部博士点基金资助项目((20110075110003);上海市教委科研创新项目(12ZS58);上海市浦江人才计划(12PJC020)

摘　　要：随着产品生命周期的缩短,如何对短生命周期产品在传统和网络双渠道的需求进行有效地预测是供应链运作中的重要问题。考虑Bass模型处理小样本数据的优越性,利用Bass模型建立了基于小样本历史销售数据的短生命周期产品需求预测模型,研究了服务和购买意愿两种因素下的Bass改进扩散模型。在此基础上,利用SVM算法构建了多影响因素的短生命周期产品双渠道需求预测模型,将改进的扩散模型修正的预测值及季节指数、进店人数、信用值等引入进来,更客观地预测2个渠道的需求,并通过算例仿真,验证了基于SVM的需求预测模型的准确性。With shortening of product life-cycle, how to predict demand of perishables has become one of important issues in supply chain operations. Due to superiority of Bass model to analyse small sample data, we firstly discuss a basic demand forecast model based on small sample of historical sales data. Then we develop an improved Bass model considering service factor and purchase factor. Accordingly, we propose a multi-factor demand forecast model for the perishables in dual-channel by using SVM （Support Vector Machine） algorithm. We include factors such as the predictive values from improved Bass model, seasonal index, number of customers, credit value in this model. Numerical simulation is carried out and the results show that SVM-based multi-factor model can improve the effect of demand forecast of perishables in dual-channel supply chain.

关 键 词：双渠道 需求预测 短生命周期产品 支持向量机 

分 类 号：TP14[自动化与计算机技术—控制理论与控制工程]