商业数据的预测模型及其算法研究

以商务流通中的经济时间序列数据为对象，设计实现了经济时间序列预测系统。在预测系统中采用了多种预测模型，设计实现了包括指数平滑算法、AR(Auto Regression)算法、Holt-Winter算法、回归分析算法在内的各种统计学算法，并且把神经网络引入到时间序列预测系统的模型中，采用动态学习的BP(Back Propagation)神经网络进行训练预测，取得了很好的实用效果。神经网络的预测模型与传统的统计方法相比，在预测的精度上有了很大的提高。

Research of prediction models and arithmetic in commerce

Based on economic time series data in commerce, we have designed and realized commercial time series prediction system. In this system, we adopted many prediction models, including exponent smooth, auto regression, Holt Winter, regression analysis and many other statistics arithmetic, furthermore, we have raised a dynamitic BP neural networks arithmetic to train net, and after lots of commercial data tests, the prediction outcome is quite satisfactory. Compared with the traditional statistical methods, the model based on the neural networks arithmetic has a much higher precision.