School of Mathematics, Statistics & Computer Science


The Teaching and Learning Office of UKZN’s College of Agriculture, Engineering and Science recently hosted a four-day Data Science Hackathon event at the UNITE Building on the Howard College campus.

Hackathon 2023 was organised in conjunction with DARA Big Data and its partners in the Hack4dev project.

This project is the fruit of several years of collaboration between the Inter-University Institute for Data Intensive Astronomy (IDIA), a South African University partnership developing cloud computing for scientific research, DARA Big Data funded by the UK’s Newton Fund, The South African Radio Astronomy Observatory (SARAO) on behalf of the South African Department of Science and Innovation, and the International Astronomical Union’s Office of Astronomy for Development (OAD).

At the core of the project is data science skills development for students coupled with a strong academic science background.

The UKZN event featured presentations by data scientists from both academia and industry, tutorials in data science and machine learning, followed by a hackathon task.

Dr Boby Varghese, Director of UKZN’s Centre for Academic Success in Science and Engineering (CASSE), gave the official welcome; whilst Dr Nikhita Madhanpall (OAD/DARA Big Data) provided an overview of the Hackathon. She was followed by Professor Febe De Wet from North-West University, who spoke to the topic of Localising the Mozilla Common Voice Platform for South Africa’s official Languages. The issue of machine learning in Academia and Industry was tackled by Mr Asad Jeewa of UKZN.

Two Hackathon projects were then presented to the participants. The first project involved Natural Language Processing (NLP) and sentiment analysis of Twitter data. NLP involves giving computers the ability to understand text and spoken words similar to how humans can. ‘Participants attended tutorials which demonstrated how to collect and clean Twitter data, perform sentiment analysis using existing toolkits and finally, perform sentiment analysis using machine learning,’ explained local organiser, Dr Nirisha Haricharan.

The dataset used consisted of 2 000 tweets collected from a dataset called Sentiment140, which contains 1.6 million general tweets and their corresponding sentiment labels.

The second project involved image classification of galaxies, making use of galaxy images from the GalaxyMNIST dataset. The dataset contained 10 000 images which had been classified through the Galaxy Zoo citizen science project.

‘This dataset was used to build an image classifier model,’ explained Madhanpall. ‘Students also looked at how they could perform this classification using unsupervised learning methods in cases where they didn’t have access to labelled data.’

At the end of the week, the winners were announced: Mr Abdulrahman Abu Raas (UKZN), Mr Daelin Parmanand (UKZN), Mr Lutho Sola (UKZN) and Mr Mussa Phiri (UniZulu) were the first-placed team; whilst Ms Ayogeboh Epizitone (DUT), Mr Sfundo Khumalo (UKZN) and Ms Abiodun Ikotun (UKZN) were placed second.

Haricharan declared Hackathon 2023 a success, saying that students obtained practical experience in data science and machine learning whilst learning in a supportive and friendly environment.

Words: Sally Frost

Photograph: Supplied