A strategy for multivariate calibration based on modified single-index signal regression: Capturing explicit non-linearity and improving prediction accuracy.

Highlights: [•] A new spectral multivariate calibration method is proposed for non-linearity. [•] In quantitative analysis, the proposed method shows the best prediction performance. [•] The proposed method explicitly exhibits the type and amount of the non-linearity. [•] The proposed method has distinct adaptability for the complex spectra model. [•] The good performance by the proposed method for biochemical analysis can be expanded. [Copyright &y& Elsevier]