Social and cognitive sciences' knowledge about social behavior and social networks combined with the new computational machine learning techniques can facilitate the creation of be er models. We propose and evaluate a new methodology for nding personality traits of young adults involved in a network using hyper optimization algorithms. We used a social contagion model for the spread of behavior (measured by the physical activity level) among the participants. A part of the Big-5 questionnaire was used to gather information about people regarding their traits of openness and expressiveness. en we try to ne tune the model using machine learning algorithms. e ne tuning of questions from an intake questionnaire can be very useful in validating a model. e accuracy delivered by machine learning pure algorithms is shown to be be er, but the inclusion of data related to people's traits is bene cial in de ning their characteristics.