Serag, AEl-Zeiny, MBNassar, MWAttia, KA2019-11-282019-11-2801/05/20171386-1425https://doi.org/10.1016/j.saa.2016.07.016https://www.ncbi.nlm.nih.gov/pubmed/27423110Accession Number: WOS:000398746600016For the first time, a new variable selection method based on swarm intelligence namely firefly algorithm is coupled with three different multivariate calibration models namely, concentration residual augmented classical least squares, artificial neural network and support vector regression in UV spectral data. A comparative study between the firefly algorithm and the well-known genetic algorithm was developed. The discussion revealed the superiority of using this new powerful algorithm over the well-known genetic algorithm. Moreover, different statistical tests were performed and no significant differences were found between all the models regarding their predictabilities. This ensures that simpler and faster models were obtained without any deterioration of the quality of the calibration. (C) 2016 Elsevier B.V. All rights reserved.en-USREGRESSIONHYDROCHLORIDECHEMOMETRIC MODELSARTIFICIAL NEURAL-NETWORKSupport vector regressionConcentration residual augmented classical least squaresGenetic algorithmFirefly algorithmArticlehttps://doi.org/10.1016/j.saa.2016.07.016