Práce s daty při predikci volebních výsledků

This bachelor thesis discusses the issue of conducting pre-election polls, focusing on the standards and methodologies commonly used in the Czech Republic and the possibility of using data from social networks. The subject of the research are the processes of data collection and processing, their advantages and shortcomings. The basic research is carried out by identifying and evaluating factors that may influence the issue, followed by an analysis of relevant documents and a comparison of methodologies of selected research agencies. The factors identified include legislation, standards of supranational member organisations and the methodologies of the individual agencies themselves. The next part of the thesis focuses on monitoring the communication of candidates in the 2023 presidential election on social media and discusses its influence on the outcome of the first round of the election and pre-election polls. As a result, the thesis proposes modifications to the basic methodologies and international standards using social media data, based on the findings on conventional and alternative methods of pre-election polls. The main shortcoming of conventionally conducted pre-election surveys is the time limitation of their relevance. An electoral model is not a prediction but a statistical model of the current situation in society. By monitoring and analyzing social network data, trends can be observed that can be incorporated into conventional methodologies and standards, according to the findings in this paper.