@article { , title = {Rebalancing stocks among retail points of sale}, editor = {Roman Buil; F.-Javier Heredia; Elena Fernández}, year = {2018}, month = {19/01/2018}, type = {MSc Thesis}, address = {Faculty of Mathematics and Statistics}, abstract = {In thisThesis, a variant of the Vehicle Routing Problem is considered where products are distributed from a depot to multiple retail stores using capacitated vehicles and not only the transportation cost but also the lost sales resulting from stock-outs at each location are minimized. Given that the real demand of a particular product at some location during a given time period can only be rougthly estimated, a deterministic and dynamic solution is proposed. This solution is divided into two phases: (i) an overnight optimization determines the set of retailers assigned to each vehicle and computes the optimal routes using the available information; (ii) a dynamic model is used separately on each vehicle in order to reoptimize its path according to the new information made available during the execution of the routes. Two exact formulations and two metaheuristics are proposed for the first phase; an exact formulation is implemented for the second phase. Moreover, an end-to-end solution is developed through the implementation of a visualization tool in R Shiny, which uses MySQL databases and shell calls to AMPL to simulate the process of a whole day.}, keywords = {teaching; VRP; metaheuristics; MSc Thesis}, author = {Sandra Orozco Martín} }