Colloquium announcement

Faculty of Engineering Technology

Department Design, Production and Management
Master programme Industrial Design Engineering

As part of his / her masterassignment

Marijn Ambrosius

will hold a speech entitled:

Designing a prediction model for short-term airline ticket demand

Date04-10-2017
Time14:00
RoomVRlab

Summary

There is a growing cost price pressure within the airline industry while the market is growing. Airfares have decreased by almost 40% over the last two decades. In the meantime, demand is rising even faster, as the passenger volume has doubled. For low-cost carriers, price discrimination is the way to go to increase market share. As a result, profit margins from ticket sales are under pressure and accurate forecasts become increasingly important to estimate low level product demand changes and adjust prices subsequently.

Accurate forecasts are crucial for revenue management. Poor demand estimates lead to incorrect pricing decisions and sub-optimal product market fit, resulting in financial losses. Airlines are organized react to market demand changes as fast and accurate as possible. As they react (to a demand change that already has taken place), there is an opportunity window between the moment the demand changes and the moment the ticket fare is adjusted.

To solve this problem, this study takes a product design approach to develop and propose a forecasting methodology to anticipate on changing ticket demand. Demand is the result of various design factors of a product, which can be both endogenous or exogenous. Examples of factors that design the product are the ticket fare, number of days booked before departure, departure day of week, but also competitor fare. After identification of the set of product design factors, their relation to demand was calculated. Next, the prediction model was designed and trained on the granular level of weekly ticket sales prediction for individual flights. Finally, by controlling ticket fares the predicted demand can be fit to the inventory available.

The proposed prediction design, variables, and model led to a forecasting accuracy improvement of 10%. By applying the model, airlines can gain new insights in the flight specific product design factor configurations that affect their product demand, subsequently predict future demand given current and future design factor values, and ultimately are able to control product sales by fitting demand to supply.