Publication: Multivariate chemometrics using r in forensic classification of certain animal hairs
Loading...
Date
2025-02
Authors
Kamisan, Muhammad Zulkhairie
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Animal hairs are one of the trace evidence that could be encountered at a crime scene, though due to their minute size, they could be overlooked by forensic investigators. Conventional methodologies for analysis of animal hair samples usually are more focused on DNA and microscopy analysis of the animal hairs, which would require expertise of the forensic investigators and consume a good fraction of time to be completed. This study utilizes rapid and non-destructive analytical technique, namely Attenuated Total Reflectance Fourier Transform-Infrared (ATR-FTIR) spectroscopy accompanied by multivariate chemometrics using R in RStudio for classification of animal hair samples of seven different animals. The multivariate chemometrics used in this study are Hierarchical Cluster Analysis (HCA), Principal Component Analysis (PCA), t-Distributed Stochastic Neighbor Embedding (t-SNE), and Linear Discriminant Analysis (LDA), which were employed to generate a visual interpretation of the results. 3D plotting was implemented on PCA, t-SNE and LDA for better separation of the clusters, which illustrates a significant difference of the clusters for pig hair samples from the other type of animals. The current work demonstrated a rapid and non-destructive method for classification of animal hair samples in forensic investigation with the implementation of chemometrics using an open-source RStudio statistical software
Description
Keywords
-