Behavioral analysis of internet users during the presidential pre-election period in the USA

This diploma thesis deals with the analysis of the online political environment in the USA within a limited period of time. In particular, it is an analysis of the presidential pre-election clickstream data from the US regions, the identification of its dominant parties, the analysis of the internet activity of its inhabitants and the comparison of these results with publicly available sources and the actual election results.The main aim of the thesis is to analyze the political prevalence in a certain state of the USA at the time of the presidential election campaigns from the October 2016, compare the results with actual election results and assess the correctness of the data used. In this work, we will be also trying to find out some interesting relations between publicly available predictions, actual election results and the results of our analysis to explain the behavior of the possible voters.This thesis is divided into two main parts. The theoretical part includes the meaning and value of big data nowadays and its relationship to the business, the definition of basic terms used throughout the work, the definitions of the statistical methods used for our analysis and a brief description of the tools used. It also describes datasets and presents an overview of the related work.The aim of the practical part is to find out a relationship between users' online behavior on the internet and their actual political behavior and define the degree of this dependence. Finally, it contains the summary and discussion of our results, the evaluation of the goal fulfillment and description of benefits for the company whose data was used for analysis.