| 基于回归系数的变量筛选方法用于近红外光谱分析 |
| |
| 引用本文： | 张世芝,胡树青,张明锦.基于回归系数的变量筛选方法用于近红外光谱分析[J].计算机与应用化学,2012,29(2):227-230. |
| |
| 作者姓名： | 张世芝 胡树青 张明锦 |
| |
| 作者单位： | 1. 青海民族大学化学与生命科学学院,青海,西宁,810007
2. 青海师范大学化学系,青海,西宁,810008 |
| |
| 摘 要： | 提出了一种基于回归系数的变量逐步筛选方法。对光谱中各变量计算其回归系数后,按其绝对值由大到小将相应变量排列,采用PLS交互检验按前向选择法逐步选择最佳变量子集。用该方法对玉米和柴油近红外光谱数据进行分析,对玉米蛋白质、柴油十六烷值和粘度分别选择出了14、12以及30个最佳变量用于建模,所得预测结果均优于全谱变量建模的预测结果。可见本方法是一种有效实用的近红外光谱变量选择方法。
|
| 关 键 词： | 回归系数 变量选择 近红外光谱 |
A variable stepwise selection method based on regression coefficient and its applications in near infrared spectroscopic data analysis |
| |
| Zhang Shizhi
,
Hu Shuqing
,
Zhang Mingjin.A variable stepwise selection method based on regression coefficient and its applications in near infrared spectroscopic data analysis[J].Computers and Applied Chemistry,2012,29(2):227-230. |
| |
| Authors: | Zhang Shizhi Hu Shuqing Zhang Mingjin |
| |
| Affiliation: | 1.Department of chemistry,Qinghai university for nationalities,Xining,810007,Qinghai,China) (2.Department of chemistry,Qinghai normal university,Xining,810008,Qinghai,China) |
| |
| Abstract: | A variable selection method based on regression coefficient was proposed in this paper.The method calculated regression coefficient for each variable firstly,then the variables were sorted with the absolute value of regression coefficients in descending way,finally,the optimum variable set was determined by PLS cross validation using a forward selection strategy.The method was used for analysis of corn and diesel fuel near infrared spectroscopic data.As results,14,12 and 30 of optimum variables were selected for modeling of corn protein,cetane number and viscosity of diesel fuel,respectively,and the prediction results of the selected variables were superior to that of full spectrum variables.It can be seen that the proposed method could effectively select the optimum variables from near infrared spectroscopy. |
| |
| Keywords: | regression coefficient variable selection near infrared spectroscopy |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
