基于GP的非线性GDP预测模型的构建与应用

针对线性GDP预测模型中存在的不足,利用遗传规划(GP)方法构建了非线性GDP预测模型,并将其应用于黑龙江省的GDP预测当中。利用黑龙江省1978—2005年实际GDP数据作为建模样本,对模型进行了训练,结果表明,与灰色GM(1,1)预测模型相比,基于GP的非线性预测模型平均拟合误差只有2.86%,远远低于GM(1,1)预测模型8.26%的结果。利用构建的基于GP的非线性预测模型对2006年黑龙江省的GDP进行预测,相对预测误差只有2.74%,远远低于GM(1,1)预测模型14.68%的结果,表明GP非线性预测模型的预测效果比较理想。最后利用基于GP的非线性预测模型对黑龙江省2007—2020的GDP进行了预测,结果表明黑龙江省的GDP呈现出稳定增长态势,年均增长率将达到8.61%。

Construction and Application of Non - linear GDP Forecast Model Based on GP

GDP forecast plays an important role in establishing economical development plan.Aiming at the insufficiency of linear econometric models,this paper presented a non-linear model to forecast GDP by genetic programming(GP) and used the model to forecast the GDP of Heilongjiang province.Using the real GDP value of 1978-2005 to construct the model and compared with the forecast model based on GM(1,1) the application result indicates that the average relative error of GP is only 2.86% much lower than 8.26% of the later.Using the model to forecast the GDP value of Heilongjiang province in 2006,the relative error is only 2.74% much lower than 14.86% of GM(1,1) model which indicates the non-linear forecast model based on GP performs better.Using the non-linear forecast model based on GP to forecast the GDP of Heilongjiang province in 2007-2020,the result indicates that the real GDP value will increase steady by 8.61% per years.