A comparative study between least-squares support vector machines and partial least squares in simultaneous spectrophotometric determination of cypermethrin, permethrin and tetramethrin

Abstract

Least-squares support vector machines (LS-SVM) and partial least squares (PLS) have received considerable attention in the chemometrics for multicomponent analysis. A comparison was made between LS-SVM and PLS methods by applying them to simultaneous spectrophotometric determination of cypermethrin, permethrin and tetramethrin. The UV absorbance spectra of the methanolic solutions of the analytes were measured in the concentration ranges, 0.1-12.9, 0.1-24.9 and 0.1-13.8 µg mL-1 for cypermethrin, permethrin and tetramethrin, respectively. The simultaneous determination of cypermetrin, permetrin and tetrametrin by UV spectrophotometry is a difficult problem in analytical chemistry, due to spectra interferences. The resolution of mixture of cypermetrin, permetrin and tetrametrin by chemometrics methods was performed. The root means square error of prediction (RMSEP) for cypermetrin, permetrin and tetrametrin by PLS and LS-SVM models were 0.1629, 0.2828, 0.1984 and 0.0766, 0.0907, 0.0467, respectively. The capability of the method for the analysis of synthesis and real matrix samples were evaluated by the determination of cypermethrin, permethrin and tetramethrin with satisfactory results. The resultant data explained that LS-SVM produced better results than PLS.