Visualization of Election Data Using Feature Engineering for Result Prediction

State election results in India yield a variegated and unique set of insights. Often it is hard to grasp the nuances of local electoral dynamics, and harder to communicate it to an audience not deeply engaged in the local political process, or to an audience from outside the state. While national news outlets do a fairly good job of analyzing and communicating the federal elections, the state elections in contrast have largely remained devoid of rigorous analysis and insightful communication. In this paper, using the data from the Assembly elections that took place in the state of Punjab in 2017, we present a process based on Feature Engineering and a set of interaction design and visualization methods to present complex insights and predictions. The general principles thus derived, will not only aid analysts and journalists to present their insights more effectively, but also empower the readers, depending on their level of interest and civic engagement, to go beyond what is presented and to discover new insights for themselves. Here the main focus is on how to clean and prepare the dataset, create new features out of the existing ones and then predict the results using a popular machine learning algorithm that may lead to more efficiency in prediction.