On the morning of the 2019 federal election, Professor Bela Stantic breathed a sigh of relief as the results of his analysis hit mainstream media, via an exclusive article for news.com.au.
Not wanting to influence the outcome, the Griffith Head of School for Information and Communication Technology delayed his prediction with News Corp’s online flagship until midday on election day, when he revealed his belief that Scott Morrison would be returned to the top job, despite official polls saying otherwise.
What followed was a frenzy of media appearances, as news outlets clamoured to hear from the only expert to correctly predict the outcome.
“I’ve worked in data analytics for a long time because of my strong mathematical and data science background – it’s what I enjoy doing. I’m not good with names, but I’m very good with numbers.”
So, who is Professor Stantic?
As Head of School and the Director of Big Data and Smart Analytics lab within the Institute for Integrated and Intelligent Systems, Prof Stantic has long been Griffith’s proponent for the power of big data in analysing major event outcomes.
Attention from media as well as both national and international organisations has increasingly come his way after he correctly predicted 49 of the 50 states won by the Republicans in the 2016 US election, which ultimately brought President Donald Trump to power.
At the time of his forecast, Democrat Hilary Clinton was still expected to win the presidency.
That same year, the vote to leave the European Union after the Brexit Referendum was also prophesised by Prof Stantic.
These predictions were not made with a cursory glance at polls and random social media threads – they were formulated after days of crunching millions of Tweets, that not only analysed the figures broadly but also the underlying targets and sentiment within each one.
An interest in astrophysics and a fondness for beach running close to his Gold Coast home allows Prof Stantic some downtime away from the big data game, but he admits numbers are never far from his mind.
“Even as a child, I realised that I read numbers differently than others and I am able to easily find interesting and often hidden information.
“Later, thanks to data science and immense power of computing, I could finally lift it to unimaginable level.”
How does he do it?
Between PhD supervisions, day-to-day meetings, and the calls and emails and that fall into his academic remit, Prof Stantic is bound to be crunching the numbers from his office at Griffith’s Gold Coast campus.
Just one Tweet being worked on by one of his specially crafted algorithms can encompass an entire screen and resemble a tangle of numbers that slowly tick over.
“Twitter is public, very dynamic, and also has a vast volume of posts – that’s why it’s useful for predictions.
“But there are other social media platforms that can also be useful to gain insight into different areas and identify valuable information, which I am also harnessing in a number of funded projects in the Big Data and Smart Analytics lab at Griffith University.
“It can be a lengthy computing intensive process. In the last few days before Australia’s 2019 election I overheated my Big Data cluster – all the CPUs were working at almost 100%.
“My algorithms work in a similar logic to the human brain for reasoning. That’s how I’ve created them in order to analyse and sift data that reflects evolving sentiments.”
What is the future of big data in predictions?
Businesses are starting to realise the power of Big data analytics and many of them are venturing into it – some with more success, others with less.
Prof Stantic believes the reason for the slow uptake so far was resistance and the difficulty in “breaking some barriers”; another reason is lack of data scientists.
He believes the work he is doing at Griffith, coupled with the media attention and calls for advice from international governments and other external organisations that accompany his predictions, are helping this realisation to occur faster.
“Anything can be predicted if there is sufficient data. The issue is that prediction can influence output and some things made public too early can cause the result to change.
“Predicting elections I don’t see as a challenge any more, but I’m now focusing on fake posts and fake news, to identify where people aren’t telling the truth in their posts. It’s my lie detector for social media.”
Griffith Sciences IMPACT event
Prof Bela Stantic will present Big Data: It was not a Miracle, it was Mathematical Certainty, on Tuesday, June 18, at the QCA Lecture Theatre and Gallery Building, South Bank campus, Griffith University, 5.30-8pm. Please register via the link.