Fifty-five red hair dyes were analyzed and evaluated using multivariate statistical techniques including agglomerative hierarchical clustering (AHC), principal component analysis (PCA), and discriminant analysis (DA). The spectra were grouped into three classes, which were visually consistent with different shades of red. A two-dimensional PCA observations plot was constructed, describing 78.6 percent of the overall variance. The wavelength regions associated with the absorbance of hair and dye were highly correlated. Principal components were selected to represent 95 percent of the overall variance for analysis with DA. A classification accuracy of 89 percent was observed for the comprehensive dye set, while external validation using 20 of the dyes resulted in a prediction accuracy of 75 percent. Significant color loss from successive washing of hair samples was estimated to occur within 3 weeks of dye application. (Published Abstract) Figures and references
Downloads
Similar Publications
- Genetic mapping of 15 human X chromosomal forensic short tandem repeat (STR) loci through multi-core parallelization
- Population Genetic Analysis with the NimaGen IDseek OmniSNP Identity Informative SNP Typing Kit
- Forgotten Spaces: The Structural Disappearance of Migrants in South Texas, chapter in The Marginalized in Death: A Forensic Anthropology of Intersectional Identity in the Modern Era