Predictive Analytics on Gender Classification using Machine Learning

The present study suggests a statistical Machine learning method to predict gender based on facial features using different facial features of a person. The dataset contains eight variables, of which seven are independent, and one dependent variable and the values are in categorical format, i.e., 0's and 1's. It contains two values, i.e., either male or female. Facial features such as nose length, nose width, forehead height, forehead width, are the most important for the determination of gender. For human recognition, we analyse many features in AI, such as iris, thumbprint, faces, voice, etc., for unique identification [1]. However, over all these features, facial representation is an important metric for Gender classification. Every individual has unique face metrics, but the similarity between males and females is determined through the measures of features. Analytical factors like race, gender, and age can be extracted from facial features. Thus, a successful gender classification approach can be a base and boost the performance of many machine learning applications like human recognition, biometric verification, and intelligent human-computer interfaces. Recognising gender through facial features can become unvarying due to the high variance in the illustration and poses of the faces [2]. In this analysis process, we have implemented various machine learning algorithms concerning classification, such as Logistic classification, Neural Networks, SVM, KNN, Decision tree, Random Forest, and Boosting algorithms. We have compared the accuracy results of all the algorithms applied and finally found the in-Boosting Gradient boosting algorithm fits the dataset best with high accuracy. We have used the python programming tool for Machine learning. Based on the best-fit algorithm, we have further processed data for predictions.