缺失偏态数据下线性回归模型的统计推断

Statistical Inference for Linear Regression Model with Missing Skew-normal Data

研究缺失偏态数据下线性回归模型的参数估计问题,针对缺失偏态数据,为克服样本分布扭曲缺点和提高模型的回归系数、尺度参数和偏度参数的估计效果,提出了一种适合偏态数据下线性回归模型中缺失数据的修正回归插补方法。通过随机模拟和实例研究,并与均值插补、回归插补、随机回归插补方法比较,结果表明所提出的修正回归插补方法是有效可行的。

We investigate the estimation of regression coefficient, scale parameter and skewness parameter for liner regression model with missing skew-normal data. In order to overcome the disadvantages of sample distribution distorted, improve the effect of estimation of regression coefficient, the scale parameter and the skewness parameter, we propose a corrected regression imputation method for linear regression model with missing skew-normal data. Compared with mean imputation, regression imputation, random regression imputation methods, simulation studies and a real example show the corrected regression imputation method is useful and effective.