Election prediction using twitter and GIS

Authors:

Abstract
The 2019 Indian general election is scheduled to be held in7 phases from 11 April 2019 to 19 May 2019, millions of voters will cast their votes during the elections, the largest democratic exercise in human history. Twitter is one of the most popular social media platforms that are used to express views regarding topics like sports, movies, politics, etc. Social Media has become one of the key platforms for politicians and political parties to promote themselves in the lead up to the elections. The aim of this paper is to do sentiment analysis of the Twitter network in the months leading up to the election and build a model that will be able to interpret the political subject matter. The analysis will be based on the politicians and the parties involved in the elections. This analysis can be used to acquire knowledge of the sentiments involved in the elections and predict the outcome of the upcoming elections. The models built are a Bidirectional Recurrent Neural Network model. The model is optimized using varying neural architectures and hyper parameter tuning. Kernel Regularization methods are used to address over fitting to get the best performance.