基于六种回归模型对门急诊量的预测  

Forecasting the number of outpatient by six regression models

作　　者：高洪达 陈荷[1] 冯启明[1] 

机构地区：[1]广西医科大学信息与管理学院卫生事业管理系

出　　处：《中国现代医生》2015年第25期119-121,共3页China Modern Doctor

摘　　要：目的比较直线回归模型、对数曲线模型、二次曲线模型、三次曲线模型、幂函数曲线模型、指数曲线模型六种模型在预测广西全区医院门急诊人次数上的准确性,为决策者提供科学的参考依据。方法依据2004~2011年广西全区医院门急诊人次数数据,分别采用直线回归模型、二次曲线模型、三次曲线模型、对数曲线模型、幂函数曲线模型、指数曲线模型对2012年门急诊人次数进行预测,并选用决定系数R2、绝对误差、相对误差等指标评价模型。结果直线、二次曲线、三次曲线、对数曲线、幂函数曲线、指数曲线模型的R2分别为0.999、0.999、0.999、0.998、0.997、0.997;相对误差分别为5.5%、4.8%、5.1%、13.1%、12.2%、3.0%。结论六种模型R2均在0.99以上,拟合效果较好,从预测误差来看,指数曲线模型误差相对最小,本案宜选用指数函数曲线模型。Objective To compare the effects on forecasting the number of outpatient and emergency patients in total hospitals of Guangxi by the regression of Linear, Quadratic, Cubic, Logarithmic, Power and Exponential. Methods The regression of Linear, Quadratic, Cubic, Logarithmic, Power and Exponential were used to predict the number of outpatient and emergency patients in 2012, and R2, relative error and absolute error were used to evaluate the regression.Results The R2 of the Linear, Quadratic, Cubic, Logarithmic, Power and Exponential regression were 0.999, 0.999,0.999, 0.998, 0.997 and 0.997 respectively; the relative errors were 5.5%, 4.8%, 5.1%, 13.1%, 12.2%, 3.0% respectively. Conclusions The R2 of these six models was all higher than 0.99 which means the six models have good fit. Exponential regression has fewer errors than others, so it is a more suitable model than others in forecasting the number of outpatient and emergency patients in total hospitals.

关 键 词：回归模型 门急诊人次数 预测 

分 类 号：R197.3[医药卫生—卫生事业管理]