Econometrics students develop impressive model centred around Albert Heijn's loyalty programme

For students in the Business Analytics and Quantitative Marketing specialisation, the case study focused on improving Albert Heijn’s loyalty programme, known as Mijn Bonus Box.

Under the guidance of Dr. Kathrin Gruber, the students worked in teams of four to develop their ideas and solutions over the course of ten weeks. Throughout this period, they presented their progress and final findings to Albert Heijn. The proposed solutions and models of all teams are highly impressive and contribute to new insights which can be utilised by the AH Data Science department.

The team built a model that can generate predictive scores on how likely a customer is to purchase a given cross-sell item. Personalised cross-sell items are items that the customer has not purchased before but may still be relevant to him, where finding the relevancy of an item may pose a challenge. These scores can then be used to make recommendations or construct offers accordingly. The basis of the methodology is a neural collaborative filtering model. Specifically, the team built on the NeuMF model introduced by He et al. (2017) and extended it with a self-attention mechanism.

Looking back at a successful 10 weeks, the students are grateful for the opportunity to develop a solution and model for the challenging real-life Albert Heijn business case. They learned a lot and enjoyed presenting their findings to the Albert Heijn Data Science department.