• Council for Scientific and Industrial Research (CSIR) says that its algorithm-based prediction model can give a near-accurate picture of the final election results ahead of time.
• Vote tallying can take a few days to complete, but the CSIR algorithm can predict results from as small as five percent of votes tallied.
• In 2016, the CSIR accurately predicted the US elections results down to the number of states Donald Trump would win.

South Africa’s Council for Scientific and Industrial Research (CSIR) says that it will be taking part on Wednesday’s national elections by using its election prediction model to show the possible outcomes of the elections ahead of the tallied results.

CSIR says that it first introduced the model in the 1999 General Elections, with the 2024 Elections being the latest usage of the prediction model. The model can provide possible outcomes of voting districts ahead of tallies by algorithm-based predictions that use initial voter information to predict the behaviour of voters.

It relies on two core principles, namely: the sequence in which voting results are announced on Election Day and the analysis of voter behaviour patterns. It uses statistical and mathematical analysis to predict the election outcomes ahead of all the counting, which can take a few days to complete.

“When combined, these two principles enable the team to group voters or voting districts based on their past voting behaviour, utilising a statistical clustering method,” the council explains, as per SA News.

The CSIR says that with only five percent of the results tallied, the model achieves “a high degree of accuracy at a national level.” With the prediction becoming more accurate and more stable the more voter districts are tallied.

This means that the CSIR will have an almost accurate depiction of the final election results before the final votes are all tallied.

“It showcases how statistical clustering and some mathematical algorithms can achieve good predictions from a small sample of results. The election prediction model operates based on reducing the bias resulting from the ‘non-randomness’ of the incoming results that arise from the order in which the results are received,” said CSIR CEO Thulani Dlamini.

In 2016, CSIR used its algorithms to correctly predict that Donald Trump would beat Hilary Clinton to win the US presidential race that year. It even accurately predicted how many states Trump would win against Clinton.

The CSIR will have a live prediction portal online during the General Elections this week in South Africa, and days following which will give a good idea what the final results will be. Keep a look out on the council’s official website.

[Image – Photo by Element5 Digital on Unsplash]