摘要
偏最小二乘回归是一种多元统计数据分析方法,它能有效克服自变量间的多重相关性问题。本文简述了偏最小二乘回归的原理、如何确定最佳成分数以及模型辅助分析。通过GPS高程拟合实例证实,偏最小二乘回归受自变量相关性影响小,拟合效果较好,计算结果稳定。
Partial Least-Square （PLS） is a multivariate statistical data analysis method, it can effectively overcome the problem of multiple correlation between independent variables. In this paper, principle of the PLS, how to determine the best number of components and model-assisted analysis are introduced. After analyzing the GPS elevation fitting example, we can get that the effect of the PLS produced by correlation between independent variables is small, the fitting result is good and the calculation is stability.
出处
《工程勘察》
CSCD
2012年第8期60-62,83,共4页
Geotechnical Investigation & Surveying
基金
国家"十一五"科技支撑计划(2006BAG04B04)
关键词
偏最小二乘
高程异常
GPS高程拟合
partial least-square
elevation anomaly
GPS elevation fitting
作者简介
刘家兴（1986-），男（汉族），安徽亳州人，硕士．
