预测分析是电信行业知识管理系统中的一个重要部分 .利用数据仓库中存储的庞大信息 ,通过回归分析、指数平滑、灰色预测以及神经网络组合预测算法对电信行业各种数据信息进行合理的组织和预测计算 ,供决策者参考 .文中简要介绍电信行业时间序列预测系统的结构设计 ,算法设计以及预测结果的分析方法.

Forecast analysis plays an important role in knowledge management system of telecom enterprises. With the advantage of great amount storage of database warehouse, we use algorithms of regression analysis, gray prediction and neural network to organize and compute various data in telecom enterprises, so that help the predictors to make decisions. This paper discusses the structure of time series forecast system, algorithm design and analytic method of results.