基于ARIMA模型的呼吸系统疾病门诊量分析

Analysis of Outpatient Volume of Respiratory Diseases Based on ARIMA Model

目的构建呼吸系统疾病门诊量的自回归移动平均求和(Autoregressive Integrated Moving Average,ARIMA)模型并预测门诊量。方法根据2005年1月至2016年12月医院门诊呼吸系统疾病人次数据,采用R软件建立ARIMA模型。通过自相关函数和偏自相关函数初步识别模型,极大似然法拟合模型,残差图、残差自相关函数图、Ljung-Box检验诊断模型,样本内比较和样本外比较选择最优ARIMA模型。并以此模型评价模型预测能力及对2017年门诊呼吸系统患病人次进行预测。结果建立了ARIMA(2,1,3)(1,1,1)12乘积季节模型,门诊人次预测值与实际观察值相近,且真实值都位于预测区间内,绝对误差和相对误差的范围分别为1.67～12.78和1.9%～19.98%,平均为8.47和11.65%,预测2017年各月平均日门诊人次在82.20～101.25之间波动。结论 ARIMA模型可以较好地拟合呼吸系统疾病门诊人次并用于预测。

Objective To establish the autoregressive integrated moving average（ARIMA） model for outpatient volume of respiratory diseases and to predict the outpatient volm^ae. Method The AIIlMA model was established by II software based on the data of outpatient volume of respiratory diseases between January 2005 and December 2016. The model was initially identified by autocorrelation function and partial autocorrelation function, fitted by maximum likelihood method, and diagnosed by residual plot, residual autocorrelation function plot and Ljung-Box test. The optimal AtllMA model was chosen through the comparison inside the sample and outside the sample. The predictive ability of the optimal model was evaluated and it was used to predict the outpatient volume of respiratory diseases in 2017. Results AtlIMA （2,1,3）（1,1,1）~2 muhiplicative seasonal model was established. The predicted value of outpatient volume was close to the actual value observed, and the actual values were all within the interval of predicted values. The ranges of absolute error and relative en＇or were 1.67-12.78 and 1.9%-19.98% and the means were 8.47 and 11.65%. The predicted monthly outpatient volume of 2017 ranged