Abstract
Online Social Networks have become a significant part of our quotidian life. In this paper, we aim to provide a proof of concept of how social media data can be effectively extracted, processed and analyzed with powerful open source tools like R. Moreover, we aim to build a reliable methodology for testing and validating social trends by using social media data. We used API routines to establish the connection between R and Twitter, Deep Learning Models to estimate the demographics of the users, Logistic Regression Models to estimate the predispositions of the users, and Propensity Score Matching to build comparable data sets. After analyzing the Romanian Twitter users, the results of our inquiry show that most of them are relatively young and the percentage of males is significantly higher than the percentage of females. Moreover, our results confirm that facial appearances play an essential role in the popularity of an individual.