CAJ | 학술논문

目的：探讨ARMA模型在某县区死亡率动态分析中的应用，分析和预测其非意外死亡率的动态发展趋势。方法：对该地2008年1月-2012年9月的逐月死亡率进行ARMA模型拟合，采用2012年10-12月的实际月死亡率验证模型的预测效果。结果：MA(1)模型较好地拟合了该地区既往时间段上的死亡率序列，模型残差为白噪声（P>0.05），模型参数MA1,1=-0.470,有统计学意义（t=-4.010, P=0.000），AIC=431.719，SBC=435.907，模型数学函数式为：xt =35.07217+(1+0.4757 B）εt。预测得到的2012年10-12月死亡率为32.65/10万、35.07/10万、35.07/10万，2013年1月死亡率为35.07/10万，预测误差为12.25%。结论：ARMA模型可以较好地拟合死亡率的时间变化趋势，并用于预测未来的死亡率，是一种短期预测精度较高的预测模型。
목적：탐토ARMA모형재모현구사망솔동태분석중적응용，분석화예측기비의외사망솔적동태발전추세。방법：대해지2008년1월-2012년9월적축월사망솔진행ARMA모형의합，채용2012년10-12월적실제월사망솔험증모형적예측효과。결과：MA(1)모형교호지의합료해지구기왕시간단상적사망솔서렬，모형잔차위백조성（P>0.05），모형삼수MA1,1=-0.470,유통계학의의（t=-4.010, P=0.000），AIC=431.719，SBC=435.907，모형수학함수식위：xt =35.07217+(1+0.4757 B）εt。예측득도적2012년10-12월사망솔위32.65/10만、35.07/10만、35.07/10만，2013년1월사망솔위35.07/10만，예측오차위12.25%。결론：ARMA모형가이교호지의합사망솔적시간변화추세，병용우예측미래적사망솔，시일충단기예측정도교고적예측모형。
Objective:To explore the application of the ARMA model in dynamically analyzing and forecasting the non-accidental mortality of the region around a non-ferrous metal mine. Methods:The ARMA model was developed based on the monthly non-accidental mortality from January 2008 to September 2012, and then tested by the remaining data in 2012. Results: Model MA(1) could preferably fit the time series of the mortality from January 2008 to September 2012 ,and the predicting error was white(P>0.05). The moving average coefficient of this model was -0.470,which was statistically significant(t=-4.010,P=0.000).AIC and SBC were 431.719 and 435.907 respectively. The mathematic function was xt=35.07217+(1+0.4757 B)εt and the predicted mortality in October, November, December 2012 using this function were 32.65 per 100 thousands,35.07 per 100 thousands and 35.07 per 100 thousands, respectively, with the average relative error of 12.25%. Conclusion ARMA model could be used to fit the changes of non-accidental mortality and to forecast the future mortality of this region, which is a predicted model of high precision for short-time forecast.