Towards Emotional Intelligence: Analysis of Static Facial Features in LinkedIn Profile Pictures

In an attempt to distil information conveyed by LinkedIn users’ pictures, in addition to the usual information provided in their profiles, we explored a large public dataset of 10,610 Australian LinkedIn users. The dataset contained in excess of 50 parameters, of which 22 were dedicated to picture features. The study confirmed that K-means clustering and Principal Component Analysis (PCA) are viable techniques for the classification of users, based on facial feature extraction and analysis. Furthermore, the study demonstrated that reduction in feature dimensionality, using PCA, yielded a significant improvement in computational time and resource consumption.