A data-driven machine learning approach for human gender classification using explainable AI

Abstract

Since gender recognition contains a wealth of information about the differences between male and female characteristics, it is crucial and essential for many applications in commercial domains, such as human-computer interaction applications and computer-aided physiological or psychological analysis. There are lots of scientific studies that contain different approaches to classify gender. Apart from those, we have included Explainable Artificial Intelligence to interpret the usability of the model. In this paper, we study a gender classification dataset to predict gender from tabular data of specific human body features such as hair length, forehead length, height, nose width, and length. We have used several Machine Learning (ML) algorithms such as - Gaussian Naive Bayes (GNB), Support Vector Machine (SVM), Decision Tree (DT), and Stochastic Gradient Descent (SGD) to predict gender. After that, we performed a comparison study and implemented Lime on the best performing ML model to interpret the prediction of the model. The Gaussian Naive Bayes architecture with the dataset shows a reasonable accuracy of 98%, proving the usability of the model.