The days around May 11, 2013 were quite eventful for the
Pakistani users on Twitter, as Pakistan’s General Elections
were held on this date for the 2013-2018 tenure [9]. These
elections represented the full-term completion of one
democratically elected government and transfer of power to
another one. We observed a significant tweeting rate on this
topic starting from January 2013 uptil the election date.
In this paper, we focus on Twitter to analyze the relevant
Pakistani Election-based tweets, from January 2013 till 7th
May, 2013. Our goal is to determine the election winner based
on Predictive Analytics techniques and we selected Twitter
because of its previous use in election opinion analysis and
electoral prediction in different countries [7] [8] [12] [13] [16]
[17] (see Section II). After monitoring of tweets it was
revealed that three political parties were more significantly
tweeted as compared to others, i.e., Pakistan Tehreek-e-Insaaf
(PTI), Pakistan Muslim League Nawaz (PMLN), and
Muttahida Qaumi Movement (MQM). We categorized the
tweets as either in favor of (Pro) or against (Anti) a given
party. After carrying out the relevant tweet processing steps
(removal of unnecessary information), we employed
Predictive Analytics to learn decision trees [18] for each
political party with the Rapid Miner data mining tool. We then
employed these trees to accurately predict the election winner
based on tweets of 8th May, 9th May, 10th May and 11th May.
Although the results showed that PTI will emerge as the
election winner, the actual overall winner was PML-N while
PTI won comprehensively in the Khyber Pakhtunkhaa
province and in several other constituencies. In other words,
Twitter can have some type of a positive influence in the
electoral results, but it cannot be considered completely
representative of the voting population.