Search

Based on the theory that a large partial least squares (PLS) regression coefficient on autoscaled data indicates an important variable, a novel strategy for variable selection called iteratively variable subset optimization (IVSO) is proposed in this study. In addition, we take into consideration that the optimal number of latent variables generated by cross-validation will make a great difference to the regression coefficients and sometimes the difference can even vary by several orders of magnitude. In this work, the regression coefficients generated in every sub-model are normalized to remove the influence. In each iterative round, the regression coefficients of each variable obtained from the sub-models are summed to evaluate their importance level. A two-step procedure including weighted binary matrix sampling (WBMS) and sequential addition is employed to eliminate uninformative variables gradually and gently in a competitive way and reduce the risk of losing important variables. Thus, IVSO can achieve high stability. Investigated by using one simulated dataset and two NIR datasets, IVSO shows much better prediction ability than two other outstanding and commonly used methods, Monte Carlo uninformative variable elimination (MC-UVE) and competitive adaptive reweighted sampling (CARS). The MATLAB code for implementing IVSO is available in the ESI.

A method for rapid quantitative analysis of epimedin A, B, C and icariin in Epimedium was developed based on Fourier transform near infrared (FT-NIR) spectroscopy, and by adopting high performance liquid chromatography-diode array detection (HPLC-DAD) as the reference method. Multivariate calibrations models were built by partial least squares regression (PLSR) based on the full absorbance spectra (10 000–4000 cm−1) or only the most informative key variables selected by the competitive adaptive reweighted sampling (CARS) method. In comparison, the accuracy of the CARS-PLSR method was apparently higher than full spectrum-PLSR for four kinds of investigated flavonoids. For CARS-PLSR, the coefficients of determination (R2) for prediction were 0.8969, 0.8810, 0.9273 and 0.9325 and the root mean square errors of prediction (RMSEP) were 0.1789, 0.2572, 1.2872 and 0.3615 for epimedin A, B, C and icariin, respectively. The good performance indicates that the combination of NIR spectroscopy with CARS-PLSR is an effective method for determination of epimedin A, B, C and icariin in Epimedium with fast, economic and nondestructive advantages compared to traditional chemical methods.