A Tutorial on Opinion Polls for Elections and Marketing Research Using Five Approaches: Logistic Regression, Discriminant Analysis, Neural Networks with Factor Analysis, Wavelets with Feed-Forward Multilayer Neural Networks and Neural Networks with Monte Carlo Batch Processor

Abstract

In this work we show a briefly presentation of four approaches to opinion polls. The example we present here, is referred on exit polls which have been realized for the elections of Serres Municipal in Greece on October 22nd of 2006. The methodology can be applied in any opinion poll, not only for political opinion polls and exit polls, but also for marketing research opinion polls. The specific paper doesn't provide a theoretical background of the proposed methods but rather a computing methodology in some of the most popular and best software.