Predictive model for Brazilian presidential election based on analysis of social media

The prediction of presidential election outcome is key point of interest for politicians, electors and sponsoring companies. The 2018 Brazilian election presented a scenario with many uncertainties increasing prediction challenge. The utilization of social media as the promotion tools is another ne...

Full description

Main Authors: Silva, Guilherme, Costa, Mirele, Drummond, André, Li Weigang
Format: Trabalho
Language: Inglês
Published: Springer 2020
Subjects:
Online Access: https://repositorio.unb.br/handle/10482/37047
Tags: Add Tag
No Tags, Be the first to tag this record!
Summary: The prediction of presidential election outcome is key point of interest for politicians, electors and sponsoring companies. The 2018 Brazilian election presented a scenario with many uncertainties increasing prediction challenge. The utilization of social media as the promotion tools is another new scenario for both election and also prediction. In this paper, we present a Bayesian forecasting model based on the data from public opinion polls to predict the votes of undecided voters, about a third of the population. The migration of votes among candidates during the electoral period was also analyzed. By using the data from social media in the decision-making process, the proposed model and application show the capability to estimate the voting numbers of the main candidates with better accuracy than public opinion polls.