Building predictive models for election results in India: an application of classification trees and neural networks

Abstract

The 2002 Judgment of the Supreme Court of India paved the way for compulsory disclosure of information with respect to the background of candidates in elections. This information includes the assets and liabilities as well as criminal antecedents, if any. The general elections held in 2004 were the first set of elections after the implementation of Supreme Court ruling. Thus, a fairly large amount of data on the candidate’ background had become available for the first time. This data was used to build predictive models for forecasting the results of the Legislative Assembly elections of the state of Karnataka. Two different data mining techniques namely, classification trees and artificial neural networks were used to build the predictive models. The prediction accuracy ranged between 90 and 98 percent.