摘要
分析了区域经济发展的特性,并指出在区域经济发展预测中存在非线性强、波动性大的特点,从而导致常规的宏观经济预测手段难以取得理想的预测效果。核方法实现了数据空间与特征空间之间的非线性映射,建立在此基础上的SVR也就具备了优秀的非线性建模能力。首先对影响区域经济发展的各因素进行了相关分析,在此基础上提出并建立了基于SVR的区域经济短期预测模型,广东省江门市的应用实例说明了该模型的有效性。
Based on the analysis of the regional economic development, the forecasting of regional economy has highly non-linear and obvious fluctuant characteristic. As a result, the general forecasting methods for macro-economics become inefficient. Kernel method can obtain the non-linear mapping from data space into feature space. SVR based on kernel method, has excellent ability of solving non-linear problem. The correlation analysis of the factors which effect the development of regional economy were conducted first. Then the short term forecasting model of regional economy based on SVR was built. The validity of the model is demostrated by applying it to a city as a real example.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2005年第12期2849-2851,共3页
Journal of System Simulation
基金
广东省自然科学基金(032353)
国家自然科学基金(70471074)。
关键词
支持向量回归
区域经济
预测
核方法
Support Vector Regression（SVR）
Regional Economy
Forecasting
Kernel Method
