Challenge: Optimization of Retail Space: We have a certain number of shelves in a store and various display zones that must be represented. The task is to fill the shelves in such a way that the chosen layout is maximally effective in terms of revenue and also meets business requirements.
Solution: I wrote a library in Python that, based on sales data and store size, calculates how the retail space should be filled. It utilizes a Beam Search algorithm that, at each step, retains not just one best solution, but the top N solutions. This approach allows for diversity (as the best solution at each step does not always yield the best sequence) and maintains a reasonable breadth of search by not analyzing very poor filling options. This solution is automated, simplifying its use for business representatives.
Outcome: This solution will be used for zoning in the deployment of new stores and the redesign of old ones
