Explained variance
Assess the share of variation represented by the first components.
Supporting method / PCA
Explore variation before resolving spectra
PCA helps represent a set of spectra with a small number of components and choose which variations to examine.
A reproducible Python workflow
DataChem workflows compute PCA on a prepared matrix, reconstruct the data and produce residual diagnostics. They export explained variance, scores and loadings, with figures and a summary report.
Once configured, the workflow can be rerun across series. DataChem can adapt it to your formats and protocols; preprocessing, component count and interpretation remain scientific decisions.
Three complementary views
Scores
Examine how observations vary with experimental conditions.
Loadings
Locate spectral regions associated with the observed variation.
In the MCR-ALS example
PCA exploration guided the choice to test a two-component resolution. This choice must be assessed alongside residuals, noise, preprocessing and the consistency of the resulting profiles.
A PCA component is a mathematical direction of variation, not a pure spectrum. Two components do not automatically mean two sites or two chemical species.