Mátyás Jakab Keindl
In my thesis, I conduct a comparative analysis of two methods with different approaches, using a standard set of variables, with the goal of estimating voter turnout. The question is whether the more modern machine learning method (XGBoost) performs better for this task, or whether the well-established regression (OLS) method remains dominant. Of course, it is also possible that there is no significant difference between the two. Although, in general—even when observing the media—parliamentary elections that determine the composition of the National Assembly receive greater attention (Bódi and Bódi, 2011), from the perspective of this analysis, municipal elections are a more suitable case for me due to their recent nature (the most recent municipal election was in 2024, while the parliamentary election was in 2022). Based on this, I will model voter turnout in mayoral elections during this research. My units of observation will be data aggregated at the municipal level—as opposed to alternative data at the county or electoral district level—since this will provide me with a sufficient number of observations to meet the conditions necessary for the models to function. Furthermore, this is the most detailed level of data available to me for the variables I intend to use.
Supervisors: Károly Bozsonyi and Renáta Németh