基于最小二乘法威布尔溶出曲线参数估计中的自相关性与异方差性问题探讨  

Exploration of Autocorrelation and Heteroscedasticity in Weibull Dissolution Parameter Estimation Base on OLS

作　　者：邱家学[1] 吴晓明[1] 

机构地区：[1]中国药科大学,南京210009

出　　处：《中国药房》2013年第1期37-39,共3页China Pharmacy

摘　　要：目的:探讨药物累计溶出百分比威布尔(Weibull)参数估计中使用普通最小二乘法(OLS)的合理性。方法:选取溶出度数据,利用杜宾-瓦尔森(DW)检验法、斯皮尔曼(Spearman)等级相关系数检验法等对自相关性和异方差性进行分析。结果:根据数据的DW检验法结果、相关系数结果均表明在采用OLS估算Weibull参数时,确实可能存在自相关性与异方差性现象。结论:在使用OLS进行Weibull参数估计时一定要进行自相关性和异方差性检验,只有在不存在自相关性和异方差性时才有可能采用OLS进行Weibull参数估计。OBJECTIVE：To explore the rationality of Ordinary Least Square（OLS）used in Weibull parameter estimation of drug accumulative dissolution percentage.METHODS：DW test and Spearman＇s rank correlation test were applied for autocorrelation and heteroscedasticity of dissolution data.RESULTS：Results of DW test and related coefficient of dissolution data indicated that there were autocorrelation and heteroscedasticity phenomena when OLS was used for Weibull parameter estimation.CONCLUSIONS：The test of autocorrelation and heteroscedasticity must be paid attention when OLS is used.Only when no autocorrelation or heteroscedasticity exists,can OLS be used for Weibull parameter estimation.

关 键 词：威布尔参数 估计 最小二乘法 自相关性 异方差性 

分 类 号：R939[医药卫生—生药学]