作者: Chao Tan , Xin Qin , Menglong Li
DOI: 10.1080/00032710902993845
关键词:
摘要: An ensemble approach, based on the combination of multiple linear regressions in subspace and variable clustering therefore named VCS-MLR, was proposed for near-infrared spectroscopy (NIRS) calibration. By an experiment involving determination five components tobacco samples, it shown that VCS-MLR improved performance by 61.4, 23.3, 10.2, 20.5, 18, respectively, with respect to partial least squares regression (PLSR). The results confirmed can result a more accurate calibration model but without increase computational burden. Moreover, superiority highlighted small sample problems.