The information services company says its DataLabs concept enhances innovation and enables data scientists to help enterprises solve strategic marketing and risk management problems through advanced data analysis processes, research and development.
The DataLabs operates globally with labs in San Diego, London and São Paulo. The facilities are staffed by teams of scientists with PhDs, and applied research practitioners with expertise in advanced analytics and machine learning, as well as other advanced statistical methods.
“Globally, Experian has invested in DataLabs in the US, UK and Brazil,” says Coleman. “These divisions are run like fintech start-ups, resourced with super-skilled individuals. Their mandate is to enable collaboration with clients to solve problems.
“In SA, we have started to set up a satellite of this set-up with investment into a sandbox environment as well as employing data scientists across our data management, analytics and marketing solutions teams,” he says. “Experian will undertake investment in broadening the talent pool that are well-versed in data science, expand its computational processing ability to provide ease of access to traditional and alternative data for clients, as well as distributing the propositions and applications from the DataLabs to allow for collaborative problem identification and solving with clients.”
According to Coleman, these individual are hooked up with their counterparts in the regional DataLabs and are taking on the localisation of prototype solutions, using machine learning techniques. He points out that machine learning should be used to help unlock relationships or predictive patterns in existing data assets that would remain hidden from human intuition or exploration.
“Where the human mind struggles with the higher order dimensionality of certain problems, advances in computer science and applied mathematics, combined with the continual growth in data storage and processing technology, now make it possible to take on previously very complex problems to solve within a reasonable timeframe.”
Typical applications of machine learning that Experian will be engaging local organisations would be in the areas of fraud within the application and digital space; historical transaction insights to inform future transaction patterns; solving predictive assessment problems for thin file populations; and leveraging recommender techniques driven by social media feeds.
“In a quite regulated environment, South African organisations are looking for alternative, non-evasive ways to confirm or authenticate the identity of their clients, and machine learning techniques deployed on a rich dataset – combining organisation and third-party data – can aid in determining this with a high degree of certainty,” says Coleman.
“Companies also need to consider how they extract value from voice, text and display data as we expect a convergence of these means to drive future customer engagement.”