基于神经网络和时间序列的河南省老年人口系数的预测

基于河南统计年鉴2001—2013年老年人口系数的数据,利用BP神经网络模型对河南省老龄化指标进行预测,训练效果不理想.因此采用时间序列二次指数平滑法对老年人口系数进行预测,预测结果的相对误差均值为4.97%.为了更加精准地预测老年人口系数,采用时间序列和BP神经网络结合的模型对其进行预测,此方法解决了老年人口系数的非线性的映射关系,预测结果的相对误差基本控制在1%左右,因此这个模型是最优的,更加适合预测河南省老年人口系数.预测结果表明河南省人口老龄化趋势是逐渐上升的.

Prediction of Coefficient of Aging Population Based on Neural Network and Time Series

Based on the data of Henan statistical yearbook from 2001 to 2013,this writer forecasts the aging index in Henan Province using BP neural network model,the training effect is not ideal. So the time series of quadratic exponential smoothing is taken to forecast coefficient of aging population,the mean relative error of the prediction results is 4. 97%. In order to get a more accurate prediction of coefficient of aging population,the writer uses a combination of time series and BP neural network model. The method solves the nonlinear mapping relationship between coefficient of the elderly population,the relative error of predicted results is controlled at about1%; so this model is optimal,more suitable for prediction of coefficient of aging population in Henan Province.Prediction results show that the trend of aging population in Henan Province is rising gradually.