An Innovative Hotel Revenue Management System Amir Atiya Dept Computer Engineering Cairo University
Project Collaborators This is an ARP project, started May 2010, under the supervision of: Dr. Mohamed Saleh (PI), Head of the Decision support Department, Faculty of Computers & Informatics, Cairo University. Dr. Amir Atiya, Department of Computer Engineering, Cairo University. Several teaching assistants. In collaboration with Comsys Software Corp. (Eng. Yasser Kadry) Preliminary proof of concept was conceived during our work with the Data Mining and Computer Modelling Center of Excellence.
Overview on Hotel Revenue Management Revenue management is the science of controlling price and/or inventory to maximize revenue. By dynamically setting a price, one aims at optimizing the revenue. The suggested price should take advantage of forecasted demand.
Our Motivation Very few four-stars hotels or lower have any kind of revenue management mainly due to the cost of available software. Our GOAL: to produce a high quality prototype eventually leading to a product having a lower price tag than the competing international products. With this we hope we could penetrate the market for the four-star and three-star hotel sectors.
Market and Industry Analysis Two types of hotel applications: Property management systems (PMS): handles all reservations, room assignments, guest checkins/check-outs, billing, etc. Revenue management systems (RMS): revenue optimization. RMS is typically added on top of an existing PMS. Comsys produces a PMS product. The goal is to complement that with an RMS product, eventually reaching the international market.
Benefits INNOVATION: Innovative IT technology will find its way to a full scale commercial product COMPETITIVENESS: Complementing Comsys Software's suit of hotel management solutions SERVICE: For the hotel industry and the IT industry in Egypt; reasonably priced cutting edge revenue management IT system. Will be made available to hotels.
Hotel Revenue Management (Problem Description) Goal is to find optimal price and number of rooms for each category so as to maximize revenue. The pricing is dynamic and changes day by day, according to forecasted demand. Pricing too low, will result in smaller revenue per room. This will cause losing higher revenue from future higherpriced reservations. (lost opportunity) Pricing too high could leave more rooms un-booked. This leads to a sophisticated optimization problem that takes into account future bookings and their probabilities.
Hotel Revenue Management Components Hotel room arrivals and occupancy forecasting In order to optimize room allocation, we need to have an idea about future hotel traffic, and the demand for the rooms. Optimization of category prices and quantities: Makes use of the forecast obtained from previous step. Formulates a large scale optimization. Takes into account an overbooking strategy, group arrivals, and some other necessary features.
Hotel Room Arrival Forecasting Forecast hotel arrivals and occupancy for a certain horizon in the future. We need the probability density of the forecasts, so empirical time series approaches are not suitable. Arrivals t t+t Probability density of future arrivals
How Hotels Work? Reservations arrive a few days or weeks before intended arrival day. Cancellations: A reservation can be cancelled prior to arrival. Any reservation books a certain number of days, or length of stay. Some reservations are booked as a block (group reservations).
The Proposed Model We propose a Monte Carlo approach, consisting of two phases: Estimate the parameters of the reservations/cancellations processes. Simulate the reservations/cancellation processes forward in time for the horizon to be forecasted in a Monte Carlo fashion. Mean of Monte Carlo paths =final forecast Monte Carlo paths t t+t
The Proposed Model (Contd) We model reservations arrivals as Bernoulli trials, with a rate that is a function of time till arrival day. We model the cancellations as Bernoulli trials, with a rate that is a function of time till arrival day. These curves are estimated from the data. Reservations rate Cancellation rate Arrival day Number of days Till arrival Arrival day Number of days till arrival
Dynamic Pricing Model Dynamically sets prices for the rooms each night by optimizing expected revenue. Uses the highly sophisticated Monte Carlo simulator for forecasting demand. Captures the demand elasticity of the price. Models the relationship between the price and certain features of the reservation, such as remaining hotel capacity, time till arrival, etc.
Overbooking Overbooking means that the hotel books more rooms than available, in anticipation of some cancelations. Our system takes into account overbooking, by modeling cancelation probability. The system continuously checks and adjusts these overbooking limit recommendations.
Overall Revenue Management System
Some Lessons Learned Innovative idea. Has good market prospects. (Does the market need this product?) Has to be competitive or better than the best world-wide product out there. Partner company has to be a sector leader, or at least knowledgeable in the considered business.
The Process If you have an innovative idea, you are encouraged to apply for an ARP. The idea has to be: Innovative. Competitive. Applicable. Have good Market prospects. Have high quality research team and collaborating company. Proposal has to be: Clear and sells well the idea. Concise and to the point. Present a quantitative market study.
The Project Review takes a few months. Once starting a project, one has to go through a number of pre-specified milestones. A reviewer oversees the progress. It is a good idea to terminate each milestone with a detailed report.