Spectral regions selection to improve prediction ability of PLS models by changeable size moving window partial least squares and searching combination moving window partial least squares

Changeable size moving window partial least squares (CSMWPLS) and searching combination moving window partial least squares (SCMWPLS) are proposed to search for an optimized spectral interval and an optimized combination of spectral regions from informative regions obtained by a previously proposed spectral interval selection method, moving window partial least squares (MWPLSR) [Anal. Chem. 74 (2002) 3555]. The utilization of informative regions aims to construct better PLS models than those based on the whole spectral points. The purpose of CSMWPLS and SCMWPLS is to optimize the informative regions and their combination to further improve the prediction ability of the PLS models. The results of their application to an open-path (OP)/FT-IR spectra data set show that the proposed methods, especially SCMWPLS can find out an optimized combination, with which one can improve, often significantly, the performance of the corresponding PLS model, in terms of low prediction error, root mean square error of prediction (RMSEP) with the reasonable latent variable (LVs) number, comparing with the results obtained using whole spectra or direct combination of informative regions for a compound. Regions consisting of the combinations obtained can easily be explained by the existence of IR absorption bands in those spectral regions.     

多组分体系同时定性定量测定的逐步回归分析法

阐述了逐步回归分析法用于多组分体系同时定性定量分析的原理、实验及数据处理方法。分析了硝基酚类人工样、未知样及合成样。该法可用于仅知可能含有的组分范围的样品分析。     

Refining adsorption parameters of n-butyl alcohol at the interface of mercury electrode with aqueous NaF and Na2SO4 solutions by correcting the adsorbate concentration

As an adjunct to the regression analysis of differential capacitance curves, which allows refining the adsorption parameters, a program, which takes into account possible errors in volume concentrations of organic substances, is developed. Using this procedure, the earlier data on the differential capacitance of a mercury electrode in aqueous solutions of normal butanol (1-BuOH) containing either 0.1 M NaF or 0.05 and 0.5 M Na₂SO₄ as the supporting electrolyte are analyzed. This allows obtaining the most accurate values of adsorption parameters for the systems mentioned above within the framework of the model of two parallel capacitors and the Frumkin isotherm. It is shown that, when a linear dependence of the intermolecular interaction parameter on the electrode potential is taken into account, the standard deviation of experimental capacitance values from those calculated using the mentioned model is 6.8–8.8%, which points to very high accuracy of this phenomenological model.     

Estimation of parameters characterizing the adsorption of sodium,potassium, and rubidium cryptates at the mercury electrode/solution interface

The adsorption of complexes formed by sodium, potassium, and rubidium cations with macrobicyclic ligand (kryptofix 222 with C₁₈H₃₆N₂O₆ composition) is studied as a function of the ligand concentration on a stationary mercury drop in 0.1 M solutions of corresponding sulfates and chlorides by using the differential capacitance technique. Based on the model of two parallel capacitors supplemented by the Frumkin isotherm, the adsorption parameters of studied cryptates are estimated by using the regression analysis technique. Differential capacitance curves calculated with the parameters found are compared with experimental data. The comparison of the found adsorption parameters makes it possible to reveal the effects of the nature of included cations and specifically adsorbed supporting-electrolyte anions on the adsorption behavior of cryptates under study.__________Translated from Elektrokhimiya, Vol. 41, No. 4, 2005, pp. 475–481.Original Russian Text Copyright © 2005 by Stenina, Sviridova.     

基于SVR和k-近邻群的组合预测在QSAR中的应用

为提高定量构效关系(QSAR)研究的预测精度,发展了一种新的基于支持向量机回归(SVR)非线性筛选分子结构描述符、基于k-近邻群的非线性组合预测方法.首先以均方误差(MSE)最小为原则,以留一法通过多轮末尾淘汰实施分子结构描述符的非线性SVR汰选并给出最优核函数和相应保留描述符;其次基于待测样本与训练样本保留描述符向量的欧氏距离,以不同k-近邻群子模型双重留一法预测值反映样本集的异质性;然后基于MSE最小,以留一法通过多轮末尾淘汰实施近邻群子模型的非线性SVR汰选并给出最优核函数和相应保留子模型;最后基于保留子模型以双重留一法实施组合预测.以取代苯胺和苯酚类化合物对大型溞的QSAR实例验证表明:新方法在所有参比模型中预测精度最高,且能更精细地反映描述符与化合物毒性间的非线性关系,具结构风险最小、非线性、适于小样本,能有效克服过拟合、维数灾和局极小,非线性筛选描述符和子模型,非线性组合预测,自动选择最优核函数及其相应参数,泛化推广能力优异、预测精度高等诸多优点,在QSAR研究中有广泛应用前景.     

Simultaneous measurement of glucose and glutamine in aqueous solutions by near infrared spectroscopy

A method is described for measuring the concentrations of both glucose and glutamine in binary mixtures from near infrared (NIR) absorption spectra. Spectra are collected over the range from 5000–4000/cm (2.0–2.5μm) with a 1-mm optical path length. Glucose absorbance features at 4710, 4400, and 4300/cm and glutamine features at 4700, 4580, and 4390/cm provide the analytical information required for the measurement. Multivariate calibration models are generated by using partial least squares (PLS) regression alone and PLS regression combined with a preprocessing digital Fourier filtering step. The ideal number of PLS factors and spectral range are identified separately for each analyte. In addition, the optimum Fourier filter parameters are established for both compounds. The best overall analytical performance is obtained by combining Fourier filtering and PLS regression. Glucose measurements are established over the concentration range from 1.66–59.91 mM, with a standard error of prediction (SEP) of 0.32 mM and a mean percent error of 1.84%. Glutamine can be measured over the concentration range from 1.10–30.65 mM with a SEP of 0.75 mM and a mean percent error of 6.67%. These results demonstrate the analytical utility of NIR spectroscopy for monitoring glucose and glutamine levels in mammalian and insect cell cultures.     

Adsorption parameters of cyclobutane carboxylic acid at the interface of a mercury electrode with aqueous solutions of 0.4 M Na2SO4 + 0.1 M H2SO4

Based on the results of the regression analysis of differential capacitance curves of a mercury electrode in solutions of cyclobutane carboxylic acid (CBCA) at 25, 50, and 75°C, the adsorption parameters that enter into both the Frumkin isotherm and the model of two parallel capacitors are estimated. The free energies of adsorption found from the analysis of the temperature dependence of these parameters are incorrect due to the partial blowing out of CBCA during the deaeration of the solutions. The corresponding adsorption parameters are corrected under the assumption that the hydrophobic effect associated with the increase in the liquid-water entropy at the removal from it of CBCA molecules is the same as for n-valeric acid. The physical meaning of the squeezing-out effect is revised.     

一个多元校正的稳健诊断新方法

提出一种新的稳健诊断方法，与最小二乘估计结合进行混合物光谱中非线性点的诊断。文中探讨了该方法的性能，用计算机数字模拟及实际多组分光谱体系夺其进行检验，展示了此诊断方法在分析化学计量学中实际应用的可行性。     

增量式Huber-支持向量回归机算法研究

传统的面向支持向量回归的一次性建模算法中样本增加时,均需从头开始学习,而增量式算法可以充分利用上一阶段的学习成果。SVR的增量算法通常基于ε-不敏感损失函数,该损失函数对大的异常值比较敏感,而Huber损失函数对异常值敏感度低。所以在有噪声的情况下,Huber损失函数是比ε-不敏感损失函数更好的选择,在现实情况当中。基于此,本文提出了一种基于Huber损失函数的增量式Huber-SVR算法,该算法能够持续地将新样本信息集成到已经构建好的模型中,而不是重新建模。与增量式ε-SVR算法和增量式RBF算法相比,在对真实数据进行预测建模时,增量式Huber-SVR算法具有更高的预测精度。     

LOWLAD: a locally weightedL 1 smoothing spline algorithm with cross validated choice of smoothing parameters

The computation ofL ₁ smoothing splines on large data sets is often desirable, but computationally infeasible. A locally weighted, LAD smoothing spline based smoother is suggested, and preliminary results will be discussed. Specifically, one can seek smoothing splines in the spacesW _m (D), with [0, 1] ⁿ D. We assume data of the formy _i =f(t _i )+ _i ,i=1,..., N with {t _i } _i=1 ^N D, the _i are errors withE( _i )=0, andf is assumed to be inW _m . An LAD smoothing spline is the solution,s , of the following optimization problem