Search

Variable (wavelength or feature) selection techniques have become a critical step for the analysis of datasets with high number of variables and relatively few samples. In this study, a novel variable selection strategy, variable combination population analysis (VCPA), was proposed. This strategy consists of two crucial procedures. First, the exponentially decreasing function (EDF), which is the simple and effective principle of ‘survival of the fittest’ from Darwin’s natural evolution theory, is employed to determine the number of variables to keep and continuously shrink the variable space. Second, in each EDF run, binary matrix sampling (BMS) strategy that gives each variable the same chance to be selected and generates different variable combinations, is used to produce a population of subsets to construct a population of sub-models. Then, model population analysis (MPA) is employed to find the variable subsets with the lower root mean squares error of cross validation (RMSECV). The frequency of each variable appearing in the best 10% sub-models is computed. The higher the frequency is, the more important the variable is. The performance of the proposed procedure was investigated using three real NIR datasets. The results indicate that VCPA is a good variable selection strategy when compared with four high performing variable selection methods: genetic algorithm–partial least squares (GA–PLS), Monte Carlo uninformative variable elimination by PLS (MC-UVE-PLS), competitive adaptive reweighted sampling (CARS) and iteratively retains informative variables (IRIV). The MATLAB source code of VCPA is available for academic research on the website: http://www.mathworks.com/matlabcentral/fileexchange/authors/498750 . [ABSTRACT FROM AUTHOR]

Abstract: We packed an ionic liquid (IL)-functionalized stationary phase – based on 1-octyl-3-propylimidazolium chloride covalently bounded to silica gel – into a 3.2mm×250mm column for the simultaneous separation of acidic, basic, and neutral compounds using carbon dioxide subcritical/supercritical fluid chromatography (SFC), and examined the effects of the pressure, temperature, co-solvents, and additives on the retention behavior of the analytes. The model compounds tested for SFC separation are acetaminophen, metoprolol, fenoprofen, ibuprofen, naphthalene, and testosterone. The data indicate that hydrogen-bonding and hydrophobic interactions between the analytes and the IL-modified stationary phase seem to involve in the separation process. Simultaneous separation of acidic, basic, and neutral compounds via SFC was successful at a co-solvent content of 20% MeOH, a pressure of 110bar, and a column temperature of 35°C. The relative standard deviations of the retention times and peak areas at 50ppm were all less than 4 and 8% (n =6), respectively. [Copyright &y& Elsevier]

Evaluating the quality of herbal medicine based on the content and activity of its main components is highly beneficial. Developing an eco-friendly determination method has significant application potential. In this study, we propose a new method to simultaneously predict the total flavonoid content (TFC), xanthine oxidase inhibitory (XO) activity, and antioxidant activity (AA) of Prunus mume using near-infrared spectroscopy (NIR). Using the sodium nitrite-aluminum nitrate-sodium hydroxide colorimetric method, uric acid colorimetric method, and 2,2-diphenyl-1-picrylhydrazyl radical (DPPH) free radical scavenging activity as reference methods, we analyzed TFC, XO, and AA in 90 P. mume samples collected from different locations in China. The solid samples were subjected to NIR. By employing spectral preprocessing and optimizing spectral bands, we established a rapid prediction model for TFC, XO, and AA using partial least squares regression (PLS). To improve the model's performance and eliminate irrelevant variables, competitive adaptive reweighted sampling (CARS) was used to calculate the pretreated full spectrum. Evaluation model indicators included the root mean square error of cross-validation (RMSECV) and determination coefficient (R 2 ) values. The TFC, XO, and AA model, combining optimal spectral preprocessing and spectral bands, had RMSECV values of 0.139, 0.117, and 0.121, with R CV 2 values exceeding 0.92. The root mean square error of prediction (RMSEP) for the TFC, XO, and AA model on the prediction set was 0.301, 0.213, and 0.149, with determination coefficient (R P 2 ) values of 0.915, 0.933, and 0.926. The results showed a strong correlation between NIR with TFC, XO, and AA in P. mume. Therefore, the established model was effective, suitable for the rapid quantification of TFC, XO, and AA. The prediction method is simple and rapid, and can be extended to the study of medicinal plant content and activity. [Display omitted] • We introduce a new NIR method to predict total flavonoids, XO inhibitory, and antioxidant activities in Prunus mume. • "CARS-PLS wavelength selection surpasses full PLS for xanthine oxidase inhibitory activity prediction models." • CARS-PLS can accurately screen effective variables in NIR quantitative models. [ABSTRACT FROM AUTHOR]

In this study, a new variable selection method called bootstrapping soft shrinkage (BOSS) method is developed. It is derived from the idea of weighted bootstrap sampling (WBS) and model population analysis (MPA). The weights of variables are determined based on the absolute values of regression coefficients. WBS is applied according to the weights to generate sub-models and MPA is used to analyze the sub-models to update weights for variables. The optimization procedure follows the rule of soft shrinkage, in which less important variables are not eliminated directly but are assigned smaller weights. The algorithm runs iteratively and terminates until the number of variables reaches one. The optimal variable set with the lowest root mean squared error of cross-validation (RMSECV) is selected. The method was tested on three groups of near infrared (NIR) spectroscopic datasets, i.e. corn datasets, diesel fuels datasets and soy datasets. Three high performing variable selection methods, i.e. Monte Carlo uninformative variable elimination (MCUVE), competitive adaptive reweighted sampling (CARS) and genetic algorithm partial least squares (GA-PLS) are used for comparison. The results show that BOSS is promising with improved prediction performance. The Matlab codes for implementing BOSS are freely available on the website: http://www.mathworks.com/matlabcentral/fileexchange/52770-boss . [ABSTRACT FROM AUTHOR]