Estimation of voter movements with Markov chains

The concept of "future" is known to involve uncertainty and risk. It is highly probable that with the prediction of the future, the uncertainties would be eliminated to a great extent and the desired results would be achieved.The prediction of the future may lead the risks to be turned into opportunities thus, the opportunities to evolve into earnings. This fact is also valid for political parties. With the presumption of the future possibilities in advance, the management of the events will be easier, and thus, political parties will achieve greater success in the next elections by strengthening their management strategies. Studies to make predictions about the future with the help of Markov Chains and Hidden Markov Chains have remarkably became widespread in recent years. In this study, Markov Chains Model was used to estimate the votes of the political parties and Hidden Markov Model was used to determine hidden states underlying the reasons for voters' choice. In the study done using the Markov Chain application, it is aimed to estimate the probability of which party the voters will prefer in the future and how the voting rates of political parties will be in the long term by considering the party preferences of the voters in the previous election. As a result of the analysis of the data, it was observed that voting for the Adalet and Kalkınma Party for Adana province was the highest in the party loyalty (87.06%) and Cumhuriyet Halk Party's and Milliyetçi Hareket Party's voters would increase more in the long term than other parties. In addition, a road map on the changes that the parties should make on their current policies in order to achieve their goals is presented in the study considering equilibrium (fixed) vector. In the second part of the study, which the Hidden Markov Model was applied, the factors affecting the reasons for preference were examined. As a result it has been determined that socio-economic, religious and social characteristics have an effect on voter behavior as well as psychological tendencies and image-oriented factors such as party identity, ideology, demographic factors (age, income level, gender, occupation etc.), political candidate image, political party image, election campaigns and information sources. In this context, the aim of the research is to determine the factors underlying the party preferences of voters who vote in local elections by depending on these factors. The universe of work was limited to Adana city due to time and cost limitations. The sample group sizes were calculated considering the number of voters in Adana's districts, and face-to-face questionnaire was applied to randomly selected individuals.