摘要： 针对常用销售预测方法预测精度较低的问题,在分析、试验了常用BP神经网络及训练算法后,提出了用引入月份因子及运用贝叶斯 正则化的方法,并应用于具有淡、旺季特性产品短期销售量预测,使预测精度大幅提高。

关键词: 短期销售顶测, BP神经网络, 贝叶斯正则化方法

Abstract: This paper presents a training method to reduce the overfitting of BP neural networks.Bayesian regularization and monthly factor is combined for training a BP NN. The method is applied to a short term sales foreeasting problem that has peak and off seasons.The results demonstrate that the methodology improves the forecasting precision a lot.

中图分类号: 

• F224