How to Decode Customer Loyalty
The student team, NOT the CAT from Sunway University in Malaysia, made it their mission in the Curiosity Cup 2025 to understand how data can improve retail marketing campaigns.
Their project focused on predicting which customers would accept a Gold Membership Campaign, offering 20% off all purchases.
The team worked with over 2,000 customer records, containing demographic details, shopping behavior, and online activity.
They carefully prepared the dataset, cleaning errors and removing inconsistencies to make the analysis reliable.
Using SAS software, they then explored three approaches to prediction:
- Decision Trees – easy to visualize but less precise.
- Regression Models – offering clear explanations of what drives acceptance.
- Neural Networks – highly accurate, able to capture complex patterns.
Their findings were clear: Neural Networks delivered the highest accuracy, but the team chose Regression as their recommended model, since businesses often need transparency and actionable insights as much as raw accuracy.
The analysis revealed that factors such as income, recent purchases, and web activity play a significant role in campaign acceptance.
Congratulations to Tan Wei Xian, Yuen Hau Xuan, Page Lee, and their Faculty Advisor, Jye Ying Sia, for demonstrating that smart analytics can enhance customer engagement.
Their full paper can be seen as a PDF below the main headline of this article.
