Abstract

Abstract:

Recent conclusions drawn from current polling techniques have demonstrated their inherent inaccuracies in election prediction. Polls are often inaccurate due to the limitations of analyzing human responses, including response, sampling, and non-response bias. Their poor performance has eroded the confidence Americans have in our political processes, and left campaign teams and political candidates searching for a better way to understand their standing among voters. With a plethora of historical data on election outcome factors, B.E.E.G.U.S. presents a modern machine-learning approach to understanding the relationship between a candidate’s individual platform and projected performance in the polls. The project has improved candidates’ abilities to campaign and utilize their funds more efficiently by reducing the man-power and associated data analytic costs ingrained in today’s campaign strategies. Utilizing B.E.E.G.U.S. in conjunction with pre-existing campaign techniques allows for the generation of actionable campaign strategies capable of improving candidates’ understanding of voters’ needs.