Absa upgrades SAS predictive analytics
08 July 2003 | 4197 views | 0
South African's Absa Bank is upgrading its SAS data mining and business intelligence software after recording a nine-fold return on its original investment.
The bank initially deployed SAS' analytical and Enterprise Miner packages to provide staff with a single consistent view of its more than nine million accounts.
The information system features over three terabytes of data in the warehouse, which can be used to determine factors such as customer lifetime value and to develop churn reduction techniques. The bank says two complementary architectures - Business Relevant Intelligence Architecture Nirvana (Brian) and Logical Analytical and Reporting Architecture (Lara)- are currently being being put in place to achieve this.
Dave Donkin, group executive, e-business and information management, Absa Bank, comments: "Because Absa has access to all customer information in one centralised data warehouse, we have been able to develop a focused action plan to continue improving service delivery to customers as well as streamline internal processes and systems."
The data warehouse has also enabled Absa to develop a security prediction model, with the information being used to predict the branches that have a high-risk potential for robberies. As a result, Donkin says Absa has "moved from being the most targeted of the big four banking institutions to having the lowest cash loss rate."