Computational assessment of the validity of elections

Abstract

In the concluded presidential elections of 2016, the opposition claimed that there was voter bribery, harassment of opposition candidates, un-equal media coverage, late delivery of voting materials in opposition strongholds, ballot box stuffing and table voting in Kiruhura district and cattle corridor of Uganda.The opposition came out and challenged the outcome of the Elections that they were fraudulent. This project used regression to validate claims of the opposition. The Project collected Presidential Election results data from 2001 up to 2016. 2001,2006 and 2011 data was used as training set and 2016 data was used as testing set. Three Regression techniques i.e Cross Validation,Decision Tree and Linear Model were used to compute predicted results for 2016 and later compute R score value and P value. The project found out that there was a large difference between the obtained and predicted results for Mr Museveni and Dr Kizza Besigye in Northern Uganda and for Kizza Besigye in Western Uganda. The Project also found the P values in Northern and Western Uganda to be very big for Dr. Kizza Besigye and his R value to be small in the same regions. The R Score value and P value suggest discrepancies in the tallying of Dr Kizza Besigye’s results in Northern and Western Uganda.