A carregar...
Publicação
Machine Learning Applications for Electoral Predictions: The Turkish Case
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.01 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Current presidential election electoral polls have produced more inaccurate predictions for Turkey's upcoming presidential election in May 2023. Despite difficult economic times and two significant earthquakes in 2023, the Justice and Development Party (AKP), the country's long-standing ruling party, and its leader Recep Tayyip Erdogan emerged victorious in the most recent presidential and general elections in Turkey, defying polls that predicted otherwise. In light of economic, sociological, religious, and political elements at the subprovincial level, this thesis explores machine learning (ML) algorithms to more accurately forecast electoral outcomes using the results of the most recent presidential election and a unique data set from Turkey. ML applications such as Linear Regression, Random Forest Regression, Gradient Boosting Regression, Support Vector Machine Regression, k-Nearest Neighborhoods Regression, Elastic Net Regression, and General Regression Neural Network were then applied after a distinct data set was gathered from official institutions and non-traditional data sources. According to the results, ElasticNet Regression and Linear Regression perform better than the other techniques in terms of 0.85 RMSE and 0.99 R2. Additionally, the technique allows for highly accurate spatial forecasts to be made at the sub-provincial level by researchers.
Descrição
Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies
Palavras-chave
Voting Behavior Election Prediction Linear Regression Prediction ElasticNet LISA Turkey SDG 4 - Quality education SDG 8 - Decent work and economic growth SDG 10 - Reduced inequalities SDG 16 - Peace, justice and strong institutions