摘要
用HP滤波法将我国GDP序列分解成趋势序列和循环序列。对循环序列建立ARIMA模型并进行预测,由GDP序列减去预测的循环序列得到新的趋势序列并对其建立ARIMA-ARCH模型并进行预测。将预测得到的循环序列和趋势序列作为自变量对GDP序列建立回归模型并进行预测分析。与直接对GDP序列进行建模的预测结果进行比较,该方法的预测结果精度更高,从而验证了该方法的可行性。
The sequence of our country's GDP sequence is decomposed into trend sequence and cyclic sequence by Hodrick-Prescott(HP)Filter.Set a ARIMA model for cyclic sequence and get the predicted cyclic sequence.The new trend sequence was obtained by subtracting the predicted cyclic sequence from the GDP sequence.Set a ARIMA-ARCH model for the new trend sequence and get the predicted trend sequence.The predicted cyclic sequence and predicted trend sequence are used as the independent variables to establish a binary regression model of the GDP sequence,and the final prediction model is obtained.Comparing the final prediction results with the prediction results which come from directly s etting model for the w hole GDP sequence.The experimental res ults show that the method can get a better prediction accuracy,which proves the fe asibility of this method.
作者
王丹
冯长焕
WANG Dan;FENG Chang-huan(School of Mathematics and Information,China West Normal University,Nanchong,637000,China)
出处
《福建江夏学院学报》
2018年第1期1-7,17,共8页
Journal of Fujian Jiangxia University
基金
西华师范大学基本科研业务费专项资金资助(14C004)
南充市社科规划一般规划(NC2013B027)
