时间序列与BP神经网络在新疆GDP预测中的对比分析  

The Comparison Analysis between Time Series and BP Network in Predicting the GDP of Xinjiang

作　　者：李贤锦[1] 胡锡健[1] 杨玉琴[1] 

机构地区：[1]新疆大学数学与系统科学学院,乌鲁木齐830046

出　　处：《科技和产业》2010年第5期104-107,共4页Science Technology and Industry

摘　　要：根据新疆维吾尔自治区2009年统计年鉴中的数据,构建并选用合适的时间序列模型、BP神经网络模型,对2004年-2008年的新疆GDP进行预测,并用预测结果与实际值求得相对误差,然后将两种模型的误差进行比较。结果表明,BP模型应用于新疆GDP预测较时间序列预测有较高的预测精度和良好的泛化能力。According to the data from the statistical yearbook of Xinjiang , we predict the GDP of Xiniiang between 2004 and 2008 by constructing suitable time series model,BP neural network model , and get relative error through predicted results and real data, then compare the error of time series model with the error of BP neural network model. The result shows that the predicted precise degree of BP neural network model is better than that of time series model and BP neural network model can be expandable. BP miodel applied to Gop love cast is of high precision are good gonerdition ahility thean：time senes.

关 键 词：新疆GDP 时间序列 ARIMA模型 BP神经网络 

分 类 号：F201[经济管理—国民经济]