The Modelling & Innovations team within BCB Analytics is responsible for driving FNB’s big data and digitization strategy within FNB Business via the application and implementation of predictive analytics and the provision of data-driven insights to the FNB Business Executive Committee. Our analytical efforts touch on an extremely wide variety of topics, where very little is out of scope, and include projects such as the design of fraud and anti-money laundering models for business, next-generation behavioural and application scorecards, attrition modelling, the optimization of limit allocation matrices, income and expense predictions for business, industry segmentation models and spend classification, to name a few. Furthermore, we also process day-to-day ad hoc queries with regards to analyses on business customers / accounts or data management requirements.

As part of the Customer Value Analytics team within FNB's Dynamic Decisioning department, we are responsible for adding business value across various departments within the bank by the development and implementation of innovative analytical techniques. In particular, my role predominately related to the design of predictive models and machine learning applications that enhance the informed decision making capabilities of our stakeholders. Business problems for which solutions were designed include propensity-to-take-up models, indebtedness predictions, cross-selling / up-selling models, attrition models, industry classification machine learning, transaction analysis via text analytics and analytical insights directed towards the FNB app. Stakeholders included FNB’s Debt Review Centre, FNB Connect (MVNO), FNB Digital Banking, FNB Credit Card and FNB Value Banking Solutions (Consumer). I was also responsible for analytical work carried out on FNB’s behavioral risk scoring aggregates and scorecards. Additionally, I actively provided guidance and assistance to quantitative analysts within the department who were fairly new and inexperienced in the area of machine learning and predictive analytics.

During my tenure at XDS, I was a member of both the Analytics Department’s telecommunications scorecard team and the Data Science Department. For the telecommunications team, my responsibilities included the design, development, testing and documentation of a telecoms provider's generation two in-store scorecard as well as extracting and compiling book-view analyses on the credit behavior of the telecommunications industry’s customer base. Additionally, I was also involved in the establishment of a credit strategy for the in-store approval process of a telecoms service provider under the guidance of a senior consultant. As part of the Data Science team, I designed and tested supplementary bureau attributes that were based on a customer’s behavior. This allowed external stakeholders who use bureau data for credit scoring purposes to include additional information beyond conventional payment profiles in their credit strategies.

During my time at FNB's Core Lending Products arm of the bank's Business Banking division, I was responsible for the design and intitial testing of an automated analytical process or system which could accurately classify the SIC code or Standard Industry Classification code of an SME (small and medium enterprise), with minimal manual user-defined inputs, as prescribed by CIPC. The purpose of the project was to clean up and rectify FNB's existing SIC code database, but also to identify the SIC codes of new Business Banking customers with increased precision so as to prevent the problem from recurring in the future. The entirety of this project served as the topic for my Masters degree thesis. I am very grateful for having received this opportunity from FNB, where I gained incredible exposure to the day-to-day operations of a business banking unit - especially with respect to unsecured lending to small organisations and the execution of various marketing and targeting campaigns.