Schedule delay impacts on air-travel itinerary demand

This paper examines air-travel itinerary share using aggregate multinomial logit models. The primary focus of this work is to assess the effectiveness of representing time of day preference by a continuous function and finding a preferred specification for the utility penalty of schedule delay, the difference between preferred and itinerary departure time. All other specification elements have been previously demonstrated. A base model representing time of day preference as a continuous function, using a weighted set of sin and cos curves is shown to reject representing time of day preference by a discrete function defined by time periods. An enhanced model, incorporating a penalty function for schedule delay which is non-linear increasing at an increasing rate over the first two hours and increasing at a decreasing rate thereafter, is found to be behaviorally and statistically superior to the base sin-cos model. This model is further differentiated between outbound and inbound passengers who are demonstrated to have very different time of day preferences.