African data science education programmes expand to meet diverse needs

As enterprises around the world race for an edge against competitors by using  data they have gathered to optimize business processes and create new revenue streams, institutions of higher education across Africa have begun offering data science programmes designed to produce graduates who can solve African problems and move into the enterpise world. 

Data science is a broad term that overlaps with different disciplines but broadly refers to the act of extracting value and actionable information from data, using a mix of traditional statistics and newer programming methodologies.

The term has been in use for several decades but has gained currency as enterprises have become awash in a sea of data, funneled into databases by a wide variety of new customer-facing applications, particularly mobile apps. Data science, though, can be applied to small data sets as well as the “big data” generated by mass-scale applications.

One of the newest data science programmes in sub-Saharan Africa is the School for Data Science and Computational Thinking at Stellenbosch University in South Africa, opened in July 2019 with the goal of “blazing new trails in what is still largely uncharted territory,” according to Wim Delva, who was acting director at the school at the time. The programme highlights a multidisciplinary approach embracing subjects including mathematics, computer science, mathematical statistics and AI.

The need for robust data science education in Africa was underscored by South African Minister of Higher Education Blade Nzimande during his address at the official opening of the Stellenbosch programme when he remarked that, “Data science as an academic discipline was pushed by the need for teams of people to analyze the big data that corporations and governments are collecting. The task for both government and universities is to prepare the youth, and adults, for the skills of the future.”

New programmes arise throughout Africa

It’s not just South Africa where data science is playing a more prominent role in higher education. For example, the African Center of Excellence in Data Science in Rwanda at the University of Rwanda and the AI & Data Science Research Group at Makerere University in Uganda are offering specialised programmes at both the undergraduate and graduate level.

The Rwandan programme offers PhD and Masters programmes with the aim of providing a research hub, stimulating collaboration between academia, government and the private sector.

“Our focus has been around building computational/AI methods and tools to improve efficiency or to compensate for a lack of resources in health, agriculture, transportation and so on,” said Engineer Bainomugisha, the chair of the programme and associate professor of computer science at Makerere University. “For example, we are using  artificial Intelligence and data science for human and plant disease detection, air quality monitoring and analysis and traffic analysis.  Since 2009 we have been running a special track on AI and data science in our MSc programme which has helped build local capacity in AI and DS.”  

The Makerere program is small, with space for 15 researchers, but its output and focus on developing-world issues is producing some very interesting research, including computational prediction of famine and mobile monitoring of crop disease.

University programmes, though, are not the whole solution to the shortage of data scientists, according to Delva, who recently left the Stellenbosch University program in order to develop a new, industry-led hybrid model for data science education. 

He recognized that many academics don’t have a practical background, so while they are expert at teaching concepts, their focus is not really on delivering graduates who can add value to commercial organizations. 

Some programmes try hybrid models

The Decision Science Accelerator, his new project at BluNova — a subsidiary of Blue Label Telecoms — aims to fill that gap with a hybrid model, combining aspects of a corporate outreach programme and an accelerator lab that allows students or recent graduates to develop projects. The programme, which he describes as “a triage tool for recruitment” seeks to strike a balance between the conceptual and the practical.

Decision science and data science may be seen as separate but complementary disciplines. Essentially, while data science derives insights from data, decision science applies data-based insights to recommendations for decision makers.

 

Wim Delva/BluNova

“You need … to get your hands dirty, building, breaking, and fixing something. While doing that you need access to mentors who come from both business and academia,” Delva said.

There are no tuition fees for BluNova’s accelerator program. Rather, it is funded by Blue Label Telecoms. The accelerator is not standing as competition to the data science university programs — in fact, the prgramme wants to cooperate with them, using their laboratories, recruiting graduates into the program and working with them to improve both teaching and curricula.

“We think of it as a bridging program between what you can do when you graduate and how you can learn to add real value to an organization,” Delva said.

Cape Town’s Explore Data Science Academy is operating with a similar mindset to BluNova, although its primary offering is a one-year, full-time undergraduate program. The Explore program is built around high-intensity, project-based sprints with a strong emphasis on teamwork, communication and collaboration. 

Building pipelines to the corporate world

Explore has found a way to work with corporations to build a pipeline of talent by identifying gaps in a company’s data science capabilities, and then tailoring a skills  programme to fill that gap and create an employment pipeline for that company.

For example, in a case study on its website, the Explore academy illustrates how it was given the task to build a data science programme for a large telecommunications organisation that needed to bring digital skills into its business. The academy created a programme and recruited 100 students for it.

The Explore academy includes a strong online component, particularly for students who can not study on campus. In general, data science lends itself particularly well to being taught through online learning, and data science programmes in Africa typically make extensive use of online resources. Thriving in a virtual classroom environment requires the same creative, solution-oriented mindset that characterises the best data scientists.

Most programme content — from managing and analysing data, writing code and deploying software applications – happens using cloud-based technology, which has gone a long way toward democratizing higher education and levelling the playing field for students in countries that have fewer resources than the U.S. 

Throughout Africa, there are a wide range of online courses that offer data science training at very little cost. For example, the University of Cape Town offers an eight-week online program with modules that include “Data Science with Python,” “Neural Networks” and “Hierarchical Clustering.” Kenya’s Ulearn Systems offers online data science courses on topics such as “Statistical Inference,” “Agile Project Management,” “Data Extraction.”

Many of the best programs on the continent, such as GetSmarter’s “Data Science with Python Online Course” offer a hybrid model of online courses, mixed with project-based group work and well-mentored internships that add depth and value to the experience.

DataScienceAfrica.org, meanwhile, aims to be an online hub for data science studies on the continent, offering resources, upcoming events and a directory of important players in the field. The site is well-organized but still feels a bit thin. 

Some African issues are unique to continent

Data science is a global discipline and it would be a mistake to think that only Africans can solve African problems. Some problems are unique to Africa, however.

“I think the African data science challenge is more challenging for a number of reasons; the supply of people you can work with is thinner, the data itself is lower in quality and consistency, more erratic,” Delva said. “You need to be a deeper problem solver, in order to succeed.”

South Africa, for example, poses an unusual challenge due to the massive gap between rich and poor in that country. So even though people may be living in close proximity to each other, their lived experiences are often worlds apart. In other words, it would be difficult to create a successful application for a squatter camp without data that is specific to people in that camp.

Ultimately, data science needs to be thought about as an important component of a much broader chain. For real, lasting success, Delva said, big questions need to be asked of any application: Is it scalable when you start talking about billions of daily observations continent-wide? How will it produce revenue over the long term? Does it truly address the root cause of the problem for customers? 

These questions explain why it’s so important for data science programmes on the continent to take a holistic view of the field and deliver a broad, inclusive curricula that provide a common language across all aspects of data science.

There is “a need for programs that help people to broaden their language and their skill set,” Delva said. “You don’t need to know everything but you need to understand how a UX person or a data engineer thinks and speaks and vice versa. If we understand what other people’s domains entail then we can build something that makes sense for all of us.”

Credit: Jeremy Daniel, Computerworld