Description du projet :
By 2018, 40% of B2B digital commerce sites will dynamically configure product pricing using price optimization algorithms. Adapting product prices dynamically as demand, transportation pathway costs, and predicted product success fluctuates over time, contributes to increasing revenue margin. Considering the increasing demand for dynamic price optimization algorithms, OrchardAi aims at developing a global, business-agnostic yet customisable framework for dynamic price optimization. Preliminary research initiated by OrchardAi shed light on the use of reinforcement learning for dynamic price optimization.
Results: A proof of concept of the reinforcement learning approach proposed, was implemented and packaged as a library. The proof of concept used dynamic programming to find the optimal the commission at each time period based on expected cumulative reward. A web-based parametrisable simulation tool was also provided to allow OrchardAi to experiment with the proof of concept and data provided.
Research team within HES-SO:
Ingram Sandy, Goetschi Damien
Partenaires professionnels: Hamilton-Smith Robert, OrchardAI
Durée du projet: