Two principal components capture most loudness-category variability

Evidence

A PCA of the combined NH and HL data revealed that the first two eigenvectors were sufficient for capturing at least 90 % of the variability in the listener parameters . Vector weightings for creating the MCPF catalog were evenly sampled from the range (2 standard deviations) of the individual listener weightings. Permutations of 66 sampled weightings were combined to create the 1460 MCPFs that constitute the catalog. — Dijk, Pim van, Physiology, Psychoacoustics and…, 2557

Categorization

• XREF: PCA dimensionality reduction of perceptual parameters mirrors similar eigen-decompositions across the hearing/psychoacoustics literature; connects to discussion of individual variability and its low-dimensional structure.
• corroborated: 2 independent passages support this claim — Dijk, Pim van, Physiology, Psychoacoustics and…, 2557