CSIR to use data science tool to predict US election outcome

The Council for Scientific and Industrial Research (CSIR) thinks it can predict who will win the US election, from the moment early results are posted after the polls have closed on Tuesday’s highly anticipated vote.

The government-owned South African technology research agency said on Tuesday that it has an election prediction model that uses data science to forecast electoral outcomes. It plans to put the tool to the test.

“The CSIR’s election prediction model relies on two core principles relating to voting behaviour of the electorate and the order in which voting results are declared on the night of the Election Day,” it explained in a statement.

“These two principles combined allow the CSIR team to group voters (or rather voting districts) together based on their past voting behaviour (using a statistical clustering method) and to then expect that any changes to voting behaviour in the new election will be fairly similar within each group.”

In doing so, the CSIR said it believes it can use science and technology to predict the election outcome “long before the results are officially declared”.

“Voters do not randomly allocate their electoral preferences but are influenced by political, socioeconomic and demographic factors, as well as past voting history,” the research council explained.

‘Window of opportunity’

“The CSIR’s election prediction model provides the window of opportunity for the model to provide insightful information for political analysts and the public to ponder on while they await the final result,” said Paul Mokilane, CSIR acting research group leader for data science.

The original model was developed by the CSIR for South African elections and was applied during the 1999, 2004, 2009, 2014 general elections and the 2000, 2006, 2011 and 2016 municipal elections.

“The success of the model in predicting the final result with a small error margin (usually with a percentage point away from final result) for the South African elections, including the 2016 municipal elections, led to an interest in testing the model outside South Africa, in different electoral systems,” said Mokilane.

In 2016, the CSIR model correctly predicted that President Donald Trump would win, against expectations. The CSIR plans to provide live forecasting of the US elections on its website from the moment early results become available. — (c) 2020 NewsCentral Media