Improved Estimation of Earth Rotation Parameters Using the Adaptive Ridge Regression |
| |
Authors: | Huang Chengli Jin Wenjing |
| |
Institution: | (1) Shanghai Astronomical Observatory, Chinese Academy of Sinica; National Astronomical Observatories, Chinese Academy of Sinica, 80 Nandan Road, Shanghai, 200030, P.R. China |
| |
Abstract: | The multicollinearity among regression variables is a common phenomenon in the reduction of astronomical data. The phenomenon
of multicollinearity and the diagnostic factors are introduced first. As a remedy, a new method, called adaptive ridge regression
(ARR), which is an improved method of choosing the departure constant θ in ridge regression, is suggested and applied in a
case that the Earth orientation parameters (EOP) are determined by lunar laser ranging (LLR). It is pointed out, via a diagnosis,
the variance inflation factors (VIFs), that there exists serious multicollinearity among the regression variables. It is shown
that the ARR method is effective in reducing the multicollinearity and makes the regression coefficients more stable than
that of using ordinary least squares estimation (LS), especially when there is serious multicollinearity.
This revised version was published online in July 2006 with corrections to the Cover Date. |
| |
Keywords: | Adaptive ridge regression astrometry data analysis Earth rotation parameters lunar laser ranging |
本文献已被 SpringerLink 等数据库收录! |
|