Comparison of the Quantitative Models for Predicting Gender Using Fingerprint Ridge Counts

• NCJ Number: 242795
• Journal: Journal of Forensic Identification Volume: 63 Issue: 3 Dated: May/June 2013 Pages: 320-331
• Author(s): Vandna Jowaheer; Deepika Pardassee; Arun Kumar Agnihotri
• Date Published: June 2013
• Length: 12 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This study used fingerprint ridge counts for predicting gender.

Abstract

In this study, three binary response predictive models (i.e., discriminant, logistic, and classification tree) were developed and evaluated for identifying gender using ridge counts of the fingers pertaining to the Indo-Mauritian population. The classification tree model was ranked best and can be readily used by practitioners. The fingerprints of only three indicator fingers (i.e., index, middle, and thumb) were used. The correct prediction probability of the classification tree model was 0.94. Those of the discriminant model and the logistic model were 0.92 and 0.90, respectively. (Published Abstract)