Prediction of gas chromatographic response factors by the PLS method

Tetrahedron Computer Methodology Pub Date : 1989-01-01 DOI:10.1016/0898-5529(89)90026-2

Giuseppe Musumarra ∗, Danila Pisano, Alan R. Katritzky ∗, Andrzej R. Lapucha, Franz J. Luxem, Ramiah Murugan, Michael Siskin ∗, Glen Brons

{"title":"Prediction of gas chromatographic response factors by the PLS method","authors":"Giuseppe Musumarra ∗, Danila Pisano, Alan R. Katritzky ∗, Andrzej R. Lapucha, Franz J. Luxem, Ramiah Murugan, Michael Siskin ∗, Glen Brons","doi":"10.1016/0898-5529(89)90026-2","DOIUrl":null,"url":null,"abstract":"<div><p>“Dietz” response factors (RF) have been measured under standard conditions for 100 substituted benzenes and pyridines. The data have been treated by the partial least squares (PLS) method, using as explanatory variables the molecular weight together with structural features such as the numbers of atoms of each element and of multiple bonds, functional groups, etc. The “Dietz” RF are explained to 84% of the variance by three PLS components. It is shown that Dietz RF can be predicted from the structural formulae for many classes of compounds with an average deviation of 0.05 within the model and 0.09 out of the model. This should be of considerable utility in the quantitative analysis of complex product mixtures by GC/MS, especially for those cases where some or all of the products are unavailable. The PREDICT.EXE program for the PC and Fortran code, PREDICT.FOR, as well as the raw data set used to derive the model are included on disk.</p></div>","PeriodicalId":101214,"journal":{"name":"Tetrahedron Computer Methodology","volume":"2 1","pages":"Pages 17-36"},"PeriodicalIF":0.0000,"publicationDate":"1989-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1016/0898-5529(89)90026-2","citationCount":"21","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Tetrahedron Computer Methodology","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/0898552989900262","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 21

Abstract

“Dietz” response factors (RF) have been measured under standard conditions for 100 substituted benzenes and pyridines. The data have been treated by the partial least squares (PLS) method, using as explanatory variables the molecular weight together with structural features such as the numbers of atoms of each element and of multiple bonds, functional groups, etc. The “Dietz” RF are explained to 84% of the variance by three PLS components. It is shown that Dietz RF can be predicted from the structural formulae for many classes of compounds with an average deviation of 0.05 within the model and 0.09 out of the model. This should be of considerable utility in the quantitative analysis of complex product mixtures by GC/MS, especially for those cases where some or all of the products are unavailable. The PREDICT.EXE program for the PC and Fortran code, PREDICT.FOR, as well as the raw data set used to derive the model are included on disk.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

PLS法预测气相色谱响应因子

在标准条件下测定了100种取代苯和吡啶的“Dietz”响应因子。用偏最小二乘(PLS)方法对数据进行了处理，使用分子量和结构特征(如每个元素的原子数和多个键、官能团等)作为解释变量。“Dietz”RF由三个PLS分量解释为84%的方差。结果表明，根据结构公式可以预测多种化合物的Dietz RF，模型内平均偏差为0.05，模型外平均偏差为0.09。这在GC/MS对复杂产品混合物的定量分析中应该是相当有用的，特别是在一些或全部产品不可用的情况下。该PREDICT. exe程序为PC和Fortran代码，PREDICT. exe。磁盘中包含FOR以及用于导出模型的原始数据集。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助