Multivariate (O)PLS regression methods in forensic dating

Abstract

In recent years, the use of chemometric tools together with analytical techniques has been increasing in forensics due to their useful role in extracting additional reliable and objective information, which can be of great help in finding and understanding the connections between the forensic evidence and the suspects involved in complex crimes. Although there are various reviews available that deal with discussing the most widely used chemometric methods and their applications in forensics, most of them focus on multivariate classification and discrimination methods, overlooking multivariate regression methods. The present review aims to prove the powerful applicability of multivariate regression methods in one of the most challenging fields of forensics – forensic dating. Based on the authors’ own experience and recent research studies, the review has been structured in three sections, covering the benefits and drawbacks of using multivariate regression methods in forensic dating and how to overcome these drawbacks, including also a methodological proposal on how to make optimal use of these methods. This review intends to provide a solid ground for emphasising and encouraging the application of chemometric-based methodologies in the coming years, in order to provide solutions to those challenges posed by forensic dating which cannot be solve by conventional methodologies.