Modeling and Forecasting the U.S. Presidential Elections Using Learning Algorithms

Please enable javascript in your browser.

Zolghadr, Mohammad | 2015

526 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 47977 (01)
University: Sharif University of Technology
Department: Industrial Engineering
Advisor(s): Akhavan Niaki, Taghi
Abstract:
In this project, we intend to use intelligent and learning algorithms to forecast the U.S. presidential elections. First, we considered some economic and political variables in our model. Then by using stepwise regression, we found the most significant variables. After that, we used three data mining techniques on these data. In the next step, we used support vector regression and neural networks to predict the elections. Then we compared these two algorithms with each other. Eventually, we realized how strong and accurate these methods are to predict the U.S. presidential elections. We have, also, proved that using data mining techniques is beneficial to make models more accurate
Keywords:
Neural Networks ; Machine Learning ; Support Vector Regression ; Forecasting ; Presidential Election ; Learning Algorithm ; United State Presidential Election

No TOC