季节分解法和ARIMA法预测乌鲁木齐市肺结核发病趋势效果分析

发布时间：2018-04-30 06:05

本文选题：时间序列分析 + 季节分解　；参考：《军事医学》2017年04期

【摘要】：目的比较时间序列季节分解法和差分自回归滑动平均(ARIMA)法预测肺结核发病趋势的效果,为肺结核预测预警提供科学依据。方法对新疆乌鲁木齐市2005年1月至2014年12月肺结核月发病率时间序列分别构建季节分解拟合模型和ARIMA拟合模型,对2015年各月发病率分别进行预测并与实际发病率进行比较。结果乌鲁木齐市肺结核流行表现出春季高发的年度周期性。应用季节分解法构建的拟合模型中,线性模型和三次曲线模型对2015年各月发病率预测结果的平均绝对百分误差(MAPE)分别为18.75%和92.25%,线性模型预测值整体上低于实际值,三次曲线模型预测值整体上高于实际值;应用ARIMA方法构建的拟合模型为ARIMA(2,1,1)(1,1,0)_(12),对2015年各月发病率预测结果的MAPE为9.46%,整体上预测值和实际值无明显差异。结论ARIMA法较季节分解法对乌鲁木齐市肺结核发病率的预测效果更佳。
[Abstract]:Objective to compare the effect of time series seasonal decomposition method and differential autoregressive moving average (ARIMA) method in predicting the incidence trend of pulmonary tuberculosis (TB). Methods from January 2005 to December 2014, the time series of monthly incidence of pulmonary tuberculosis in Urumqi, Xinjiang, were analyzed by using the seasonal decomposition fitting model and the ARIMA fitting model, respectively. The incidence rates of each month in 2015 were predicted and compared with the actual incidence. Results the prevalence of pulmonary tuberculosis in Urumqi showed the annual periodicity of high incidence in spring. In the fitting model constructed by seasonal decomposition method, the average absolute percent error MAPE of linear model and cubic model are 18.75% and 92.25%, respectively, and the predicted value of linear model is lower than the actual value. The predicted value of cubic curve model is higher than that of real value on the whole, and the fitting model constructed by ARIMA method is Arima _ (2) ~ (1) ~ (1) ~ (1) ~ (1) ~ (1) ~ (1) / 1 ~ (1) / 1 / 1 / 1 / 1 / 1 / 10 / 10 / 12 / 12 / 1 / 12 / 12 / 12 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 / 1 Conclusion ARIMA method is more effective than seasonal decomposition method in predicting the incidence of pulmonary tuberculosis in Urumqi.
【作者单位】：军事医学科学院疾病预防控制所;新疆乌鲁木齐市疾病预防控制中心;
【基金】：全军后勤科研重大项目(AWS14R013) 全军后勤科研重点项目(BWS14C051) 新疆维吾尔自治区自然科学基金资助项目(201442137-20)
【分类号】：R521;R181.3

【相似文献】