Machine Learning × Revenue Management
Predicting Hotel Cancellations
Hotels lose revenue to last-minute cancellations, but overbooking blind gets guests walked. Predict each reservation, then price the two errors against each other.
Moving the decision threshold from 0.50 to 0.55 recovered $10,620 more modelled profit — optimising for margin, not accuracy.
Team project · MSBA 315, AUB
View project