The Golden Thread: Planning for Success in Data-Driven Research and Business
Get valuable insights and hard-won lessons in just seven quick answers!
Learn directly from Hardus Bodenstein, a recent PhD graduate from the University of the Western Cape (UWC), South Africa.
His highly awarded doctoral thesis, Flexible feature engineering using a network flow approach, explores how advanced optimization can transform one of data science’s toughest challenges.
Explore his full work here and get connected with him on LinkedIn.
1. What did you study at university, and what inspired you to choose that field?
“I studied Data Mining, which in later years evolved into what is now more commonly known as Data Science. At school I had a passion for both mathematics and programming, and I was fortunate to choose the right course at the right university.
Coming from a small town and a small school where it was relatively easy to excel, I initially found university life quite intimidating. However, I gave it everything I had and am proud to say that I not only succeeded but also truly enjoyed the experience.”
2. How did your learning journey continue, particularly regarding your PhD?
“My PhD was a very different experience. After completing my master’s degree, I started working full-time in the short-term insurance industry, building models for CRM lifecycles and pricing. During that time, I reconnected with one of my former professors, and together we began exploring the idea of a PhD, which I pursued part-time.
Being young and energetic, it felt manageable at first. Things became more demanding when I was blessed with a daughter, and her “terrible twos” arrived just as my work responsibilities increased, and my PhD deadline drew near (already extended once). The final stretch was tough, but it taught me invaluable lessons about resilience and perseverance. Beyond the academic knowledge, I learned a lot about myself and my ability to keep pushing forward despite challenging obstacles.”
3. What led you to select your specific thesis topic, and why did you decide to use SAS software for your solution?
“In the insurance industry, I spent much of my time building predictive models with SAS. I realized that a large portion of model-building is devoted to feature engineering.
Scorecards, pioneered in the credit risk industry, are powerful yet explainable. A major part of scorecard development involves discretizing continuous features and collapsing levels of high-cardinality nominal predictors. This process is still largely manual and relies heavily on expert knowledge, particularly when business constraints must be respected during simplification. With today’s data volumes, a modeler may evaluate thousands of predictors, making the process even more complex.
Given my background in mathematical optimization, I saw an opportunity to tackle this problem optimally. I use SAS daily in my work, and I am aware of the power of the optimization tools available within SAS/OR. Implementing my solution in SAS made sense because it integrated seamlessly with my model-building workflow and allowed me to leverage familiar tools while applying advanced optimization techniques.”
4. How would you describe your overall experience with SAS?
“Very positive, it’s my home language after all 😉. SAS is robust, reliable, and offers powerful statistical and optimization tools. To me, one of its biggest attractions is the quality control and thorough debugging built into its solutions, which is not something you always get with open-source alternatives.
The learning curve can be steep, but mastering SAS has been highly rewarding and has greatly improved my ability to implement complex data-driven solutions with confidence.”
5. You graduated in 2024. What were your plans after graduation, and how does your current reality compare? What are you doing now?
“Since graduating, I have continued working in the industry, and it has been wonderful to have more autonomy over my time 😊.
I am in a more managerial role and thoroughly enjoy mentoring and upskilling junior members of the team.
I also maintain strong ties with the university, as an external moderator for their master’s program, which allows me to stay connected to the academic community while contributing my professional experience.”
6. Which skills from university do you think you use most in your career, and where do you feel your studies could have better prepared you for the professional world?
“The most valuable skills I gained are critical thinking and structured problem-solving. Even if I’ve forgotten some technical details, those analytical abilities stay with you for life. The programming, statistical modeling, and optimization skills I developed have been vital to my success in practice.
However, studying part-time while working on my PhD taught me important soft skills as well! Such as time management, prioritization, and balancing multiple responsibilities.
All skills have proven just as important in my career as technical knowledge. That said, one thing the university didn’t fully prepare me for is how messy real-world data can be. It rarely behaves like the nice, neat examples in the textbooks!”
7. What is your primary advice for current students preparing their thesis using SAS or working on data-driven research?
“Plan thoroughly from the beginning. As the saying goes: If you fail to plan, you plan to fail. Make sure there’s a clear ‘golden thread’ running through your thesis so that every section contributes to the overall research story. And don’t shy away from big problems. They may be challenging, but with the right tools and structured approach, you might surprise yourself!”
Hardus Bodenstein's journey from data enthusiast to a leading researcher and manager is truly inspiring. His dedication to bridging the gap between academic research and real-world applications reflects his passion for creating meaningful impact.
Want to read more stories from academic researchers shaping the future of analytics?
Follow the hashtag #TalentSpotlight for the next feature.
