A Study on Providing Useful Information to Voting Behavior in Election Coverage by Machine Learning

Abstract

Useful information that attracts voters to politics is needed for election coverage. In recent study, machine learning that output reasons for winning or losing election by combinations of explanatory variables was proposed. In this study, we validate the usefulness of such reasons from the voters' point of view by a conjoint experiment and a questionnaire survey. The result shows that such reasons can possibly help voters choose a candidate compared with existing election coverage keywords. The result also shows that unexpected reasons can attract voters' interestingness by showing those possibilities of winning the election in the past.