肾综合征出血热发病率的小波预测模型

Model of wavelet-based analysis and forecasting for incidence rate of hemorrhagic fever with renal syndrome

摘要:
目的建立季节性水平变化趋势时间序列小波预测模型, 提高肾综合征出血热(HFRS)发病率的预测步长及精度。
方法对原始序列进行多层小波分解, 分解后的各层分别用自回归滑动平均(ARIMA)模型进行预测, 将各层的预测值合并作为原序列的最终预测值。
结果小波预测模型4步预测精度为82.45%, 而ARIMA建模的4步预测精度为67.97%。
结论用小波预测模型对水平变化趋势的HFRS作短、中期预测是有效、可行的。

Abstract:
Objective To improve the forecasting precision and the step-length of the incidence rate for Haemorrhagic Fever with Renal Syndrome(HFRS), this paper proposed the forecasting method of the seasonal non-tendency time series called wavelet forecasting model.
Methods By wavelet decomposing, each level series was forecasted by the ARIMA model.The final forecasting results were composed of these levels forecasting values.
Results The 4-step forecasting precision of wavelet forecasting model and ARIMA model was 82.45% and 67.97% respectively.
Conclusion Wavelet forecasting model was effective and feasible for the seasonal non-tendency HFRS's incidence rate prediction in the short and the middle term.