13 December 2024
- Students accessed a dataset of 14 million sale transactions
- Continuous mentorship by engineers from the Shoprite Group
- Access to large datasets and the infrastructure needed to handle such data
A long-standing partnership between Stellenbosch University (SU) and the Shoprite Group has enabled the first cohort of graduates in the Data Science programme to engage with real life business challenges from the retail industry.
For their final year project in the Computer Science Division, for example, Günther Tonitz and Steffan Schoonbee tackled the hard question of promotional timing: When is the right time to promote which item, and for how long?
“We quickly realised the much more difficult question was whether items on promotion actually lead to success from a retail perspective,” explains Günther. In other words, if a customer buys five of the same items on promotion – what is known as pantry stocking – it may mean that they are not going to buy that product again for the next five months.
In another scenario, called off-target shopping, a customer may buy a promotional item, such as a soft drink, and then add a few bags of chips (that are not on promotion) for throwing a party, Steffan adds.
Which approach then, can be defined as a success?
To answer this question, the students were given access to two years’ worth of data from one of the Shoprite Group’s supermarkets. It included 16 000 articles, equating to 78 000 promotional combinations. This dataset was further enhanced with stock movement data over a nine-month period, covering 18 000 articles. The dataset contained over 1 million stock movements.
Working with 14 million sales transactions, Steffan and Gunther built a dashboard to predict the success of future promotions, based on the performance of previous promotions.
According to Debbie Vigario, Head of Engineering at the Shoprite Group’s Data Analytics Office (and a SU alumnus), their approach is to provide students with real life business challenges: “We support them through the analytical development life cycle while challenging their thinking along the way. We assigned mentors to students and ensured that we engaged regularly to support and guide them.”
With the launch of the B-degree in Data Science and the BEng Data Engineering programmes in 2019, the Group engaged with the various faculty leadership on ways they could partner with SU on project work. This engagement led to the various real-life projects for final-year students in the Computer Science focal area within the BDatSci programme.
Dr. Mkhuseli Ngxande from SU’s Computer Science Division says the BDatSci programme with a Computer Science focus emphasises technical expertise in software engineering, web development, and data science:
“It equips students with practical skills to develop end-to-end solutions that are both innovative and well documented. This focal area is distinct from others like Economics and Management Sciences or AgriSciences, as it centres on technical implementation, computational problem-solving, and creating software-driven data solutions rather than applying data science concepts to domain-specific problems in business, agriculture, or environmental studies.”
- Dr. Mkhuseli Ngxande from Stellenbosch University's (SU) Computer Science Division
While this is the first cohort of five BDatSci graduates in the Computer Science focal area, there are another 20 to 25 students in the undergraduate pipeline.
He says it is challenging for students to gain access to large datasets independently due to privacy concerns, licensing restrictions, and the infrastructure needed to handle such data. Collaborations like those with Shoprite are invaluable in bridging this gap.
Prof. Ingrid Rewitzky, head of the Department of Mathematical Sciences and Vice-Dean: Teaching and Learning for the Faculty of Science, says the partnership with Shoprite contributes towards “enhancing student success through unlocking opportunities for work-integrated learning and for equipping students with skills and competencies to be competitive for their future employment applications”.
The final-year BDatSci (CS) students are Dirk Hoffmann, Steffan Schoonbee, Shriyan Singh, Günther Tonitz, and Franco Uys.