Title:

The application of GM (1, 1) grey forecasting model and ARIMA model in the fitting of the mortality of the cardiovascular disease in Shandong province

作者：

赵晶¹, 郭晓雷², 吴炳义¹, 王姗¹, 王在翔¹

赵晶¹，郭晓雷²，吴炳义¹，王姗¹，王在翔¹

Author(s):

ZHAO Jing*, GUO Xiao-lei, WU Bing-yi, WANG Shan, WANG Zai-xiang

ZHAO Jing*, GUO Xiao-lei, WU Bing-yi, WANG Shan, WANG Zai-xiang

单位:

1. 潍坊医学院，山东 潍坊 261053；
2.山东省疾病预防控制中心，山东 济南 250012

Unit:

*Weifang Medical University, Weifang, Shandong 261053, China

关键词:

GM（1; 1）；ARIMA模型；心脑血管疾病；死亡率

GM（1,1）；ARIMA模型；心脑血管疾病；死亡率

Keywords:

GM (1; 1) grey; ARIMA model; Cardiovascular disease; Mortality

分类号:

R181.2

文献标识码:

A

摘要:

目的 运用GM(1,1)灰色模型和ARIMA模型对山东省心脑血管疾病死亡率进行拟合，对拟合结果行进比较。为心脑血管疾病预防提供科学依据。 方法 利用山东省全人群监测点2002-2014年心脑血管疾病死亡率数据分别建立GM(1,1)灰色模型和ARIMA模型，对建立的模型进行拟合，同时运用该模型对2015-2017年心脑血管疾病死亡率进行预测。两种模型拟合评价指标为误差平方和（SSE）、平均绝对百分误差(MAPE)两个指标。 结果 心脑血管疾病死亡率GM(1,1)灰色模型和ARIMA模型SSE和MAPE分别为1236、1189和2.75%、2.73%。2015-2017年心脑血管疾病预测死亡率（1/10万）分别为340.56、349.80、359.03。 结论 心脑血管疾病死亡率呈波动性上升趋势。ARIMA模型拟合效果优于GM(1,1),模型拟合要充分考虑数据特征。

Abstract:

Objective The aim of this study was to use GM (1,1) grey model and ARIMA model in the fitting of cardiovascular disease mortality in Shandong province to compare the fitting results, in order to provide scientific basis for the prevention of cardiovascular disease. Methods GM (1,1) grey model and ARIMA model were respectively set up using the cardiovascular disease mortality data of the whole crowd monitoring stations from 2002 to 2014 in Shandong province to establish the model fitting. At the same time, the model was used to predict the cardiovascular disease mortality from 2015 to 2017. The two kinds of model fitting evaluation index were error sum of squares (SSE), mean absolute percentage error (MAPE) two indicators. Results The GM (1,1) grey model and ARIMA model SSE and MAPE of cardiovascular disease mortality were respectively 1 236 and 1 189, as well as 2.75% and 2.73%. The prediction rates of the mortality of the cardiovascular disease from 2015 to 2017 (1/100 thousand) were respectively 340.56, 349.80, and 359.03. Conclusion The mortality of the cardiovascular disease was fluctuating in an upward trend. The effect of ARIMA fitting model was better than that of GM (1,1) fitting model. It′s suggested to pay full consideration to the characteristics of data.

参考文献
/References:

备注/Memo:

基金项目：“健康山东”重大社会风险预测与治理协同创新中心资助课题(XT1405003)；山东省医药卫生科技发展计划项目(2014WS0468)。
作者简介：赵晶(1990-)，男，研究方向：流行病与卫生统计学
通讯作者：王在翔，E-mail：WANGZX1@126.com