Stacked partial least squares regression analysis for spectral calibration and prediction

Abstract

a Two novel algorithms which employ the idea of stacked generalization or stacked regression, stacked partial least squares (SPLS) and stacked moving-window partial least squares (SMWPLS) are reported in the present paper. The new algorithms establish parallel, conventional PLS models based on all intervals of a set of spectra to take advantage of the information from the whole spectrum by incorporating parallel models in a way to emphasize intervals highly related to the target property. It is theoretically and experimentally illustrated that the predictive ability of these two stacked methods combining all subsets or intervals of the whole spectrum is never poorer than that of a PLS model based only on the best interval. These two stacking algorithms generate more parsimonious regression models with better predictive power than conventional PLS, and perform best when the spectral information is neither isolated to a single, small region, nor spread uniformly over the response. A simulation data set is employed in this work not only to demonstrate this improvement, but also to demonstrate that stacked regressions have the potential capability of predicting property information from an outlier spectrum in the prediction set. Moisture, oil, protein and starch in Cargill corn samples have been successfully predicted by these new algorithms, as well as hydroxyl number for different instruments of terpolymer samples including and excluding an outlier spectrum.

Genetic algorithm‐based wavelength selection method for spectral calibration

“…Ni et al . applied this concept to spectral calibration 36. They proposed stacked PLS (SPLS) and stacked MWPLS (SMWPLS) methods based on iPLS and MWPLS, respectively.…”

Stacked PLS for calibration transfer without standards

“…A full description of stacked PLS regression has been reported elsewhere [14,18]; only a brief summary is given here. The stacking emphasizes the local features of the multivariate spectral response to a target property in the resulting regression model.…”

“…Improved robustness for prediction of the external property has been demonstrated for prediction sets taken on the same instrument used to collect calibration data [18]. It is of interest here to determine whether that robustness is maintained in the application of a stacked calibration model to a dataset taken on a secondary instrument.…”

“…PLS calibration was based on a modified version of an algorithm from the PLS Toolbox version 2.1.1 from Eigenvector Research. Software implementing SPLS regression, OSC and FIR preprocessing and MUP was written in-house in the Matlab language following published algorithms [11,13,18]. Prior to building PLS and SPLS models, all data were centered using the mean of the calibration set of sediment spectra measured on the primary instrument.…”

Stacked Regression With a Generalization of the Moore-Penrose Pseudoinverse

“…There are also some new papers which propose extensions to SR (Džeroski and Ženko (2004), Rooney et al (2004a), Rooney et al (2004b), Xu et al (2007), Ozay and Vural (2008), Ni et al (2009)), Ledezma et al (2010), Shunmugapriya and Kanmani (2013). An informative overview of SR methods can be found in Sesmero et al (2015).…”