Five case studies across telecom, business intelligence, machine learning, credit risk and market strategy, each with the challenge, the approach and the measurable outcome.
Unsupervised segmentation of 100,000 banking customers into six risk profiles, pairing K-means with a DBSCAN outlier pass that isolates the highest-risk 3.3% for manual review.
Every project here follows the same honest structure: the commercial challenge, the analytical approach, and the outcome with real numbers attached. Where work is covered by confidentiality, I describe the method and the magnitude, never invented specifics.
Want the full detail?
I’m happy to walk through any of these projects: the data models, the campaign logic and what I’d do differently next time.