基于SVR模型的重庆市生态安全指标预测模型研究

Study on Prediction Model of Ecological Security Index in Chongqing City Based on SVR Model

鉴于实际工作中生态安全指标数据统计的滞后性,通过采集并整理重庆市1988年至2007年与生态安全指数极度相关的28项指标构建了重庆市生态安全指标体系,并以此作为训练样本数据,以2008、2009年对应指标作为输出结果,分析比较多变量灰色预测模型、径向基神经网络、支持向量回归方法,通过对比真实统计数据做出误差分析,得到支持向量回归重庆市生态安全指标预测模型。该模型对重庆市主要生态安全指标进行预测的结果具有明显的优势,可以用于实际预测。

In actual practice, owing to hysteresis of the conventional statistical analysis on ecological safety index, this article implemented Multivariable Grey model, Radical Basis Function Network and Support Vector Regression to input extremely relevant samples of ecological safety index from 1998 to 2007 in Chongqing. The outputs generated by the three models were evaluated and compared with ecological safety index gathered in 2008 and 2009. According to the er- ror analysis between the outputs and the actual index, more accurate predictions were produced by the Support Vector Regression model. In conclusion, the Support Vector Regression model is applicable to actual practice and has higher ac- curacy than the other two models.