The project is about understanding the impact of the corona virus on voting behavior in Germany. The variables include citizens' risk perception, evaluation of political measures, trust in the federal government, and media consumption. Data is publicly available at https://search.gesis.org/research_data/ZA5667?doi=10.4232/1.13520) The multi-class analysis includes EDA, and finding the best predictive model for the choice of political party.
Feature importances indicated by the SHAP value reveal that political orientation is still the most influential feature regardless of the COVID outbreak. The citizens' perceptions on the government policies, the COVID measures and media exposure are mainly shaped by the political orientation, which in return influence the voting behavior. Interestingly, the COVID-related features have a higher influence on far-right (Afd) or far-left(Die Linke) tparty choices than the political parties at the center (CDU/CSU).
The ROC-Curve indicates the overall performance of the classification model at all classes.