Textile fibers obtained from crime scenes can provide trace evidence to help solve criminal cases through forensic science. Examining the samples involves the use of different analytical procedures to determine associations between different fabric fragments. The current study explored multivariate statistical methods applied in synthetic fiber forensic science comparisons. The literature review of the current research identified various multivariate approaches such as Principal Component Analysis (PCA) and Partial Least Square Discriminant Analysis (PLS-DA) to analyze spectroscopic and chromatographic data. The capstone project used Fourier Transform Infrared (FT-IR) Spectroscopy instrumental methods for the analysis of standard synthetic fibers. Afterwards, multivariate data analysis and data mining techniques using Aspen Unscrambler Software were applied to analyze the spectral data of four types of fiber obtained from the FT-IR analysis. Data were first subjected to preprocessing methods including Savitzky-Golay first derivative method and Standard Normal Variate (SNV) method. Then, PCA model was generated in which the four types of fiber were separated based on wavenumber properties. Followingly, PCA Projection and Soft Independent Modelling by Class Analogy (SIMCA) were built as classification models. The classification models classified the test samples accurately, which indicated that the PCA model generated in this research could be used as a method of classification for fiber samples.
Professional Studies (MS)
Department, Program, or Center
Graduate Programs & Research (Dubai)
Aljanaahi, Abdulrahman, "Multivariate Statistical Analysis Applied to the Forensic Analysis of Synthetic Fibers" (2021). Thesis. Rochester Institute of Technology. Accessed from