Inter-university virtual hackathon unites African data scientists

Read time 2min 30sec
Almost 2 500 machine-learning solutions were submitted in 7.5 hours for the #UmojaHackAfrica.
Almost 2 500 machine-learning solutions were submitted in 7.5 hours for the #UmojaHackAfrica.

The University of Cape Town and Stellenbosch University were among the three universities that scooped the top prizes in Zindi Africa’s first Pan-African inter-university hackathon, #UmojaHackAfrica.

Over a thousand data scientists virtually joined hands across closed borders to participate in the artificial intelligence online hackathon.

While this event took place in extraordinary circumstances, with the world in the grip of the coronavirus (COVID-19) pandemic, students from 70 universities in 15 African countries competed for cash prizes, and seized the opportunity to collaborate and hone their machine-learning skills.

Zindi Africa works with firms, non-profit organisations and government institutions to develop and host data-driven challenges. The company says it now has 12 000 registered data scientists on its platform.

Student teams at universities competed in various online challenges, while honing their skills and developing solutions to complex real-world problems.

The virtual hackathon ended with Stellenbosch University student Geoffrey Frost scooping the prize for the Hotspot Fire Prediction challenge. Frost developed a fire outbreak prediction algorithm using data collected in the Democratic Republic of Congo.

The Xente Purchase Prediction challenge saw Team Statistically Significant – Ivan Jericevich, Ehsaan Jeraak, Patrick Chang and Taru Singhal from the University of Cape Town – take the prize for their solution to the challenge of predicting purchasing behaviour for e-commerce customers in Uganda.

Firas Baba, from the Higher School of Communication of Tunis, Tunisia, won the prize for the SAEON Marine Invertebrate Classification challenge. His solution uses computer vision methods to classify marine invertebrates from images taken in South African waters.

The winning teams in each of the three challenges will win a share of $2 000, as well as a share of $15 000 for their university.

“I am thrilled to see that despite the challenges faced by the restrictions of the coronavirus, we were able to demonstrate just what is possible through online and remote platforms like Zindi,” says Celina Lee, Zindi chief executive officer.

“More so, I am impressed by the quality of outcomes and that so many data scientists across Africa showed up to test their skills at #UmojaHackAfrica.”

Almost 2 500 machine-learning solutions were submitted in 7.5 hours.

Runner-up prizes were awarded to: National School of Electronics and Telecoms of Sfax (Tunisia), University of Lagos (Nigeria), Federal University of Technology Akure (Nigeria), Nelson Mandela Advanced Institute of Science & Technology (Tanzania), Multimedia University of Kenya (Kenya) and Moringa School (Kenya).

In addition to the main prizes, Zindi says it awarded $17 000 and around 60 licences for LinkedIn Learning in prizes on the day of the competition, testament to the belief that sponsors and communities have in Zindi and Africa’s growing AI community.

The challenge sponsors and partners included Microsoft, African Bank, Google Research, Liquid Telecom, Alliance4ai, Instadeep and the Field Institute.

See also