Interpretable Dimension Reduction and Clustering for Complex Chromatographic Datasets

The standard response is to reach for a dimension reduction technique. Principal component analysis (PCA) is the most familiar—well-documented, widely available, and easy to plot. PCA finds linear combinations of the original variables that capture the largest share of variance in the data, and the resulting two- or three-dimensional score plots can reveal groupings that would never be obvious from the raw numbers.