Past deze baan bij jou?
Wat ga je doen?
Within the Loyalty, Analytics & Media Services department of Albert Heijn the Customer Solutions & Insights (CSI) team works on analyzing large amounts of data to improve our understanding of consumer behavior and to build personalized digital services. This is realized by obtaining insights into marketing campaign effects, building of customer segmentation models, estimating customer lifetime value, providing personal recommendations, etc. Solutions are realized by building algorithms with the help of effective data science/machine learning techniques (e.g. Probabilistic Graphical Models, Deep Belief Networks, Gradient Boosting, Random Forests, etc.) and state-of-the-art infrastructure (Microsoft Azure). Therefore, CSI can deliver, test, implement relevant models in multi-disciplinary teams to ensure AH stays on the frontline of the most favorable & healthiest supermarket of the Netherlands.
To guarantee that models can provide business value the data needs to be reliable, timely and of high quality. This applies to internal data as well as external data. Our goal is to improve our data landscape through standardization and automation such that data is readily available. Moreover, significant steps are taken to further improve the usefulness of existing data (such as the Bonuskaart data) through innovation and optimization. The goal is to further improve the data analysis process such that we can explain consumer behavior more effectively and provide personalized services to create the best and unique customer shopping experience. The CSI team plays a center role to achieve this goal.
AH uses various channels to interact with the customer to provide relevant services, such as on-/offline product recommendation, e-mailed personalized Bonus offers, meal inspiration, do-not-forget hints etc. To improve the customer experience over these touchpoints none of the interactions must be driven in isolation. Instead, every interaction should be dependent on previous interactions with the customer to optimize the next action (i.e. ‘Next best action’). Moreover, all customer interactions must be orchestrated considering contextual information, customer profile, state of the customer, etc.
Effective orchestration comes down to solving a sequential decision making problem where a certain utility function needs to be optimized. The optimization of the utility function should result in a policy that defines what the next best action is given a certain state of the orchestrated system. Also the utility function can be used to control the behavior of the orchestration providing a certain level of control over the decision making process.
One important aspect in evaluating the algorithm is to consider customer behavior. To evaluate the effect of the orchestration on customer behavior the algorithm will be tested live on the ah.nl website using different learned policies.
What do we offer?
We expect the student to:
- Understand AH’s possible customer interactions over touchpoints;
- Review/study literature on decision making/machine learning techniques to solve sequential decision making problems (e.g. (Partially Observable) Markov Decision Process, Influence Diagrams, Monte-Carlo tree search, etc.);
- Make an action plan on which methods and data to use to perform orchestration;
- Implement and test solutions to orchestration;
- Perform live evaluation of the performance on conversion rate of the different proposed methods through A/B testing;
- Document code, report and present findings to internal stakeholders
- Affinity with retail domain;
- Interest in applying machine learning/decision making techniques to do orchestration;
- Excellent programming skills in Python and SQL, and
- The student is required to be enrolled in a MSc program Artificial Intelligence, Computer Science, Applied Mathematics, Operation Research, Statistics or related during the entire duration of the graduation project.
Provide your motivation letter and C.V. For more information contact dr. Patrick de Oude (firstname.lastname@example.org), Senior Data Scientist and Data Science Lead at AH Consumer Solutions & Insights.