摘要
本文介绍了最小一乘准则下的直线回归方法,由于最小一乘准则下的回归之稳健性较最小二乘准则好,当回归模型中的随机误差不服从正态分布,特别是在怀疑数据中有异常数据时,最小一乘准则下的回归具有较好的统计性能。
This paper introduced a robust method of regression on the basis of the Least Absolutecriterion. The least absolute regression is more sensitive than least squares when the error inthe regression model departs from the normal distribution, especially when there are outliersin the observations. Therefore, using least absolute regression we can obtain a stable statistical character.
出处
《现代预防医学》
CAS
1992年第4期207-209,共3页
Modern Preventive Medicine
关键词
直线回归
最小一乘准则
Linear regression Least absolute criterion Robust
