摘要
近代回归分析中的数据诊断理论和方法在测量数据处理中可以得到应用，测量中的有关误差理论也可用于近代回归分析，甚至扩展其数据诊断范围。本文首先阐明数据诊断理论与可靠性和模型误差理论间的关系，在文献〔2〕、〔4〕基础上，导出了不等权观测时强权点、奇异点和强影响点的检验和度量公式，针对实际测量数据处理问题的需要，给出了在一般情况下的Cook距离公式以及用降权法进行回归数据诊断。
:
The method of data diagnosis in recent regression analysis and the theory behind
it can be applied to the data prpcessing of observations. the relevant theories of obser-
vation errors can be applied in turn to recent regression analysis and even can extend
the range of data diagnosis. This paper first elucidates the relation between the theory
of data diagnosis and the reliability as well as the theory of error in model. On the
basis of [2] and [4], formulas for testing and measuring hish leverage case, influen-
tical case and outlier of observations with unequal weights are derived. In an attempt
to meet the need of data processing in practical measurements Cook's distance formula
in general case is given and regression data diagnosis is carried out by the method of
descending weight.
出处
《测绘学报》
EI
CSCD
北大核心
1989年第3期168-174,共7页
Acta Geodaetica et Cartographica Sinica
