He predicted Brexit, Trump’s 2016 win, and ScoMo’s 2019 “miracle” – now the data scientist modern Nostradarmus has predicted the election winner.
According to the polls, the 2022 federal election is as tight as it can possibly get.
The Labor Party, they say, holds the faintest lead over the Coalition – but we all remember what happened last time we thought Labor was going to win.
It’s only fair to be a bit sceptical or wanting a bit more certainty as to who will be running the country after May 21.
Times like these call for oracles, or Paul the Octopus, or data scientists like Bela Stantic.
The Griffith University professor not only predicted Scott Morrison’s “miracle” victory in 2019, he also tipped Donald Trump to become the 46th US President in 2016.
So we called him back to predict the winner of today’s federal election – and even he says the race will go right down to the wire.
“The situation changes every hour,” Professor Stantic told news.com.au. “Postal votes, preferences, and independents will be crucial.”
That said, he believes Anthony Albanese will become the 31st prime minister of Australia, with the Labor Party clinching a “clear win”.
Surprised? Perhaps not, given Labor is a short favourite among the bookies, and it has held the (narrowing) lead throughout the campaign, and the party’s own secret polling has Mr Albanese poised to form government.
To make the predictions, Professor Stantic analysed sentiment across social media, crunching millions of tweets through data analysis programs at Griffith University’s Big Data and Smart Analytics lab on the Gold Coast.
The programs reveal what people have been thinking about and feeling throughout the election.
The same system that predicted the US election victory and the 2019 Coalition “miracle”, also predicted the LNP’s 2016 election win and predicted, against the British polls, that the Brexit referendum would prevail.
On the back of his previous wins, and few losses, Professor Stantic said his big data research had shown analysing as little as 5 per cent of social media data equated to 95 per cent accuracy.
Basically, the numbers are pretty close to spot on.
“It is scary how accurate prediction can be done by analysing social media,” Professor Stantic said after his Trump premonition came true.
He said the old-school method of telephone polls deliver “inconsistent results” for a number of reasons. Whereas self-generated data – through social media use – is a “secret treasure of information if we know how to discover it”.
“(Data) makes predictions about the future based on the patterns of the past, and finds relationships buried in the data that no one has noticed,” he said.
Although the professor was running and re-running the numbers late into the eve of election day, the modern Nostradarmus isn’t as confident with this prediction as he was in 2019.
Why? For one, the gruelling campaign was “very influential” on making public favour see saw, he says.
And, second, an eleventh-hour surge in public favour for the Coalition and Scott Morrison has added uncertainty to his prediction.
“The campaign changed a lot on a few occasions in each direction, with a significant move toward the Coalition,” Professor Stantic said.
This week, alone, he analysed a sample size of more than 1.4 million voters and found 12 per cent had changed their voting intentions between parties and candidates more in this election than in 2019.
Professor Stantic said 20 per cent of people showed “uncertainty” in their language about who they would vote for and were still undecided in the final week of campaigning.
“I have a prediction (of) who will win, but I am not happy with the confidence,” he said, adding that the last election was “much easier” than the 2022 race.
Even so, the lab has predicted Anthony Albanese’s Labor party will win, calling time on the Coalition’s decade in power.
While it’s all yet to be seen whether our very own – very human – Paul the Octopus has done it again. But if his track record is anything to go by, we might have dropped the biggest spoiler of the entire election.
