多元光度测定病态系统的岭回归估计

考察了含多种酚类的有机混合体系的同时测定，发现对于这类化学与光谱性质皆十分相近的混合物体系，最小二乘法和卡尔曼滤波方法都不能给出令人满意的结果，甚至出现负值，而脾岭回归估计方法对此类病态体系进行分析，所得结果明显优于最小二乘法与卡尔曼滤波方法。

Ridge Regression Estimation for Ill-conditioned Systems of Multicomponent Spectrophotometric Determination

The multicomponent spectrophotometric determination of the organic mixture system containing phenolic compounds was investigated. For these ill-conditioned systems with very similar spectral characteristics, both ordinary least squares regression and Kalman Filtering methods are not able to give satisfactory results. We used the ridge regression method newly developed in statistical literature to treat these ill-conditioned systems. The results obtained are evidently superior to those obtained by ordinary least squares or Kalman methods.