The , developed by Eigenvector Research, is a professional-grade software suite designed for chemometrics and multivariate data analysis within the MATLAB environment. Since its initial release, it has become a standard in both academic research and industrial applications—particularly in fields like analytical chemistry, pharmaceuticals, and process engineering. Core Capabilities and Features
Assume you have a near-infrared (NIR) spectra matrix X (100 samples × 500 wavelengths) and a concentration matrix Y (100 samples × 2 components).
✅ – Standard and extended methods ✅ Advanced preprocessing – MSC, SNV, derivatives, wavelets, and more ✅ Variable selection – VIP, selectivity ratio, genetic algorithms ✅ Classification tools – SIMCA, PLS-DA ✅ Model diagnostics – Outlier detection, cross-validation, randomization tests ✅ Interactive graphics – Score plots, loadings, contribution plots matlab pls toolbox
The function is a standout feature: it automatically selects the optimal number of latent variables based on a user-specified criterion (e.g., minimum RMSEV or the F-test of Haaland and Thomas), iterating through cross-validation folds.
The PLS Toolbox is not merely a collection of regression scripts; it is a comprehensive environment for the entire lifecycle of multivariate data. Its capabilities can be categorized into three primary pillars: exploratory analysis, regression, and classification. MATLAB PLS Toolbox The , developed by Eigenvector
The toolbox implements rigorous validation strategies:
This process is vital for determining the optimal number of latent variables to include in the model. Including too few components results in underfitting, while including too many captures noise. Through its cross-validation interface, the PLS Toolbox helps users navigate this trade-off, ensuring the final model is robust and generalizable. It also supports test-set validation, providing a secondary check on model performance. PLS & PCR ✅ – Standard and extended
: Standard methods like Partial Least Squares (PLS), Principal Components Analysis (PCA), and Nonlinear methods like locally weighted regression.