Chemometrics Overview

SpectraSherpa focuses on practical chemometrics workflows for spectra.

Main Questions

• What structure is visible in my spectra? Use PCA.
• Can spectra predict a concentration or property? Use PLS regression.
• Can spectra assign a known class? Use KNN, PLS-DA, or SIMCA.
• Is a sample consistent with a reference class? Use SIMCA QC.
• Can overlapping mixture components be resolved? Use MCR-ALS or SIMPLISMA-style tools.
• Which peaks or regions matter? Use peak finding, variable selection, VIP, or library comparison.

Start With PCA

For most new datasets, PCA is the first serious model. It reveals clustering, outliers, drift, preprocessing effects, and whether labels or targets plausibly align with spectral variation.