Fernando Hernandez, a Senior Risk Consultant and Trainer at Palisade, was tasked with developing a model for predicting demand and price elasticity for a domestic airline catering to the tourism industry in Costa Rica. This predicting model uses neural networks technology powered by Palisade’s NeuralTools application, a sophisticated data mining application that makes new predictions based on the patterns of known data. By imitating brain functions to “learn” the structure of data, NeuralTools can take new inputs and make intelligent predictions.
The objective of this trained network was to adequately predict how many passengers would purchase tickets on a particular flight and the optimum price point, based on a wide range of condition combinations. Once the data set was created, the neural network was set to be trained and tested. Upon training and testing, a sensitivity analysis on 21 variables was performed to predict the passenger demand for a particular flight. A sensitivity analysis showed which factors held the greatest weight in determining the number of passengers per flight. The most impactful factors included mean fare, distance and competitor strength.
A data entry table was created utilizing fare increments of $5.17, starting at $50, and all the way up to $200. The model predicted that occupancy would remain more or less steady—between 17 to 23 passengers— regardless of the fare, until the $175 threshold was reached. Once this mean fare was surpassed, passenger demand abruptly decreased driving down total expected revenue for the route. At a mean fare of above $185—a mere increase of $10 per ticket—occupancy dipped to less than five passengers.
While demand and price elasticity is not an exact science, NeuralTools helped the airline utilize data it already had to determine the most profitable price point and adjust it based on the numerous factors that impact air travel.
To see the full case study, click here.