Modeling the Proximate Covariance Property of Air Travel Itineraries Along the Time-of-Day Dimension

Document Type

Journal Article

Publication Date


Subject Area

planning - route design, ridership - behaviour, organisation - competition


Travellers, Travelers, Time of day, Route (Itinerary), Ridership, Periods of the day, Patronage (Transit ridership), Ordered generalized extreme value models, Nested logit models, Level of service, Itinerary, Human behavior, Highway users, Generalized extreme value models, Departure time, Decision making, Competition, Behaviour, Behavior, Airlines, Airline industry, Air travel, Air lines, Air carriers


The estimation of advanced air travel itinerary share models formulated to capture interitinerary competition dynamics along the time-of-day dimension is reported. These models predict airline ridership at the itinerary level and aid carriers in long- and intermediate-term decision making. Previous estimations by the authors indicated increased competition among air travel itineraries within broad time periods. In this study, a more realistic time-of-day competition dynamic is modeled with the ordered generalized extreme value (OGEV) model and hybrid OGEV models. The hybrid OGEV models incorporate the traditional OGEV model structure with other GEV components such as the nested logit (NL) and weighted NL model structures. These hybrid OGEV models measure interitinerary competition along the carrier or level-of-service dimensions along with the time-of-day dimension. The estimated OGEV models have the property of proximate covariance in which itineraries that are closer to each other by departure time exhibit greater covariance and therefore greater substitution or competition with each other than with itineraries that are more separated in time. Independent variables for all of the models measure itinerary service characteristics such as level of service, connection quality, carrier attributes, aircraft type, and departure time. Finally, all models estimated in this study offer insights into air traveler behavior, with the advanced models outperforming the more basic specifications with regard to statistical tests and behavioral interpretations.