Avis Budget EMEA Ltd is a leading car rental company in Europe, Middle East and Africa. Over the last few years the company have invested considerable resources in the bespoke design and development of a Revenue Management (RM) Forecasting and Optimisation system to allow Avis to maximise the revenue generated by its fleet of vehicles.
The system is split into modules. One major module is to forecast the future customer bookings using an exponential smoothing method, with consideration of no-show rate, cancellation probability, and seasonality, each of which is computed by separate modules. Once the demand forecast is generated, they are, along with the available fleet of vehicles, passed over to an optimiser module to maximize revenue, while taking into account a large number of operational and business rules.
The optimisation problem is formulated as a mixed-integer program with demand deemed as deterministic. If stochastic demand is incorporated, the expectation is used instead and the problem becomes a probabilistic nonlinear program, which is not trivial to solve. In light of this, the expectation is approximated by a set of linear inequalities in order to reduce the non-linear problem to a mixed-integer program to which the standard Branch and Bound algorithms apply.
The main decisions from the RM system are two levers which control both the supply and demand. One of these is to limit the availability of certain products so that fleet is saved for more profitable customers. The other is to shuttle cars between rental stations, typically moving cars from areas of low demand to areas of high demand.
Dong Li is an operational research analyst at Avis Budget EMEA Ltd, which he joined in 2010. His main responsibility is to leverage operational research techniques in the car rental industry and in particular the revenue management problems.
He obtained his PhD in Management Science from Lancaster University in 2011, and was awarded the Kingsman Prize 2011 for his high quality of research and publications. He has also spent three years working at Intel Corporation on problems concerned with the planning and scheduling of semi-conductor production.