The Airline Group of the International Federation of Operations Research (AGIFORS) held a conference in October 2020 that included keynote addresses from KLM Royal Dutch Airlines and Airbus, as well ...as three panels that included representatives from 11 airlines throughout the world that focused on how COVID-19 is impacting and reshaping the airline industry. This paper presents key themes that emerged from these discussions, including the impact of border closures on airline operations and demand forecasts; the shift in development priorities within revenue management departments; and outlooks for how passenger preferences, booking curves, and fare product restrictions may change after the COVID-19 pandemic.
This is a summary of the author's PhD thesis supervised by Michael Schyns and defended on April 29, 2016 at the University of Liège, Belgium. We examine two problems as part of this dissertation. The ...first is a cargo loading problem. The second problem we examine involves the estimation of itinerary choice models that include price variables and correct for price endogeneity using a control function that uses several types of instrumental variables.
Price differentiation is a common strategy in many markets. In this paper, we study a static multiproduct price optimization problem with demand given by a discrete mixed multinomial logit model. By ...considering a mixed logit model that includes customer specific variables and parameters in the utility specification, our pricing problem reflects well the discrete choice models used in practice. To solve this pricing problem, we design an efficient iterative optimization algorithm that asymptotically converges to the optimal solution. To this end, a linear optimization (LO) problem is formulated, based on the trust-region approach, to find a “good” feasible solution and approximate the problem from below. A convex optimization problem is designed using a convexification technique to approximate the optimization problem from above. Then, using a branching method, we tighten the optimality gap. The effectiveness of our algorithm is illustrated on several cases, and compared against solvers and existing state-of-the-art methods in the literature.
•We study a static multiproduct pricing problem under discrete mixed logit model.•Utilities include individual specific variables and heterogeneous price parameters.•We design an algorithm that asymptotically converges to the optimal solution.•The effectiveness of our algorithm is illustrated on several cases.
•First itinerary choice model for all U.S. markets that accounts for price endogeneity.•Highly refined departure time of day preferences estimated.•Segmentation includes direction of travel, ...distance, time zone, day of week.•Results underscore the importance of correcting for price endogeneity.
Network planning models, which forecast the profitability of airline schedules, support many critical decisions, including equipment purchase decisions. Network planning models include an itinerary choice model that is used to allocate air total demand in a city pair to different itineraries. Multinomial logit (MNL) models are commonly used in practice and capture how individuals make trade-offs among different itinerary attributes; however, none that we are aware of account for price endogeneity. This study formulates an itinerary choice model that is consistent with those used by industry and corrects for price endogeneity using a control function that uses several types of instrumental variables. We estimate our model using a database of more than 10million passenger trips provided by the Airlines Reporting Corporation. Results based on Continental U.S. markets for May 2013 departures show that models that fail to account for price endogeneity overestimate customers’ value of time and result in biased price estimates and incorrect pricing recommendations. The size and comprehensiveness of our database allows us to estimate highly refined departure time of day preference curves that account for distance, direction of travel, number of time zones traversed, departure day of week and itinerary type (outbound, inbound or one-way). These time of day preference curves can be used by airlines, researchers, and government organizations in the evaluation of different policies such as congestion pricing.
There is growing interest in using online outsourcing platforms that are part of the “gig economy” to conduct surveys for academic research. This interest has been driven in part by the belief that ...compared to traditional survey data collection methods, internet-based marketplaces such as Amazon Mechanical Turk (MTurk) enable one to collect survey data cheaper and faster from a larger, more diverse participant pool. However, many have questioned whether models based on survey data from these online marketplaces are similar to models based on survey data from more traditional platforms. To investigate this research question, we used MTurk and Qualtrics (a traditional market research firm) to survey air travelers. Our results showed that MTurk and Qualtrics respondents had distinct socio-demographic characteristics, but we found no statistical evidence for different air trip characteristics. In our data, proportionately more MTurk respondents were in the younger, single, male, and lower-income categories than for Qualtrics respondents. We found that airline itinerary choice models estimated from the MTurk and Qualtrics survey data were similar, with the key difference related to price sensitivities. Although our results provide evidence that MTurk can be used for travel demand modeling applications, we offer words of caution for others planning to conduct surveys in online marketplaces, particularly for those seeking to recruit more than 1000 participants or for those targeting specific geographic areas.
Inspired by European actions to fight organized crimes, we develop a choice-based resource allocation model that can help policy makers to reduce the number of pickpocket attempts. In this model, the ...policy maker needs to allocate a limited budget over local and central protective resources as well as over potential pickpocket locations, while keeping in mind the thieves’ random preferences towards potential pickpocket locations. We prove that the optimal budget allocation is proportional in (i) the thieves’ sensitivity towards protective resources and (ii) the initial attractiveness of the potential pickpocket locations. On top of this, we also study two alternatives of our choice-based resource allocation model: one where pickpocket probabilities are enforced to be equal over the pickpocket locations, and one where the decision-making process of the thief becomes deterministic, with known preferences, as observed by the policy maker. For both alternatives, we also derive the optimal budget allocation and compare it with the initial budget allocation using numerical experiments. Finally, we illustrate how these optimal budget allocations perform against various others budget allocations, proposed by policy makers from the field.
•We develop a choice-based resource allocation model to fight pickpocketing.•We show how to optimally allocate resources for the setting with and without fairness constraints.•We test how the optimal solutions compare against resource allocations proposed by policy makers from the field.
•Horizontal price agreements can fall within the scope of exemptions to antitrust competition if they are expected to create pro-consumer benefits.•Inspired by these exemptions, we introduce a ...cooperative game in which a set of operators can collectively decide at what price to offer sustainable mobility services.•Numerical experiments illustrate that various well-known allocation rules (i.e., proportional rules and the Shapley value) do not always generate core allocations.•We introduce a market share exchange rule and prove that this rule always generates core allocations.
Horizontal price agreements can fall within the scope of exemptions to antitrust competition if they are expected to create pro-consumer benefits. Inspired by such horizontal agreements, we introduce a cooperative game in which a set of transport operators can collectively decide at what price to offer sustainable urban mobility services to a pool of travelers. The travelers choose amongst the mobility services according to a multinomial logit model, and the operators aim at maximizing their joint profit under a constant market share constraint. After showing that various well-known allocation rules (i.e., proportional rules and the Shapley value) do not always generate core allocations, we present a core-guaranteeing allocation rule, the market share exchange rule. This rule first allocates to each transport operator the profit he or she generates under collaboration, and then subsequently compensates those transport operators that lose part of their market share, which is paid by the ones that receive some extra market share. This exchange of market share is facilitated by a unique price, which can be expressed as the additional return by cooperating per unit of market share. Finally, we show that, under some natural conditions, the market share exchange rule still sustains the collaboration when the transport operators need to pay back part of the joint profit to society.
•New data-driven framework for enhancing choice models.•Conservation of standard DCM interpretability while increasing predictive power.•Systematic utility divided into a knowledge-driven and a ...data-driven part.•Demonstration of framework’s effectiveness on the MNL and NL models.•New choice models referred to as the Learning Multinomial Logit (L-MNL) and Learning Nested Logit (L-NL) models.•Experiments on publicly available datasets based on revealed or stated preferences.
In discrete choice modeling (DCM), model misspecifications may lead to limited predictability and biased parameter estimates. In this paper, we propose a new approach for estimating choice models in which we divide the systematic part of the utility specification into (i) a knowledge-driven part, and (ii) a data-driven one, which learns a new representation from available explanatory variables. Our formulation increases the predictive power of standard DCM without sacrificing their interpretability. We show the effectiveness of our formulation by augmenting the utility specification of the Multinomial Logit (MNL) and the Nested Logit (NL) models with a new non-linear representation arising from a Neural Network (NN), leading to new choice models referred to as the Learning Multinomial Logit (L-MNL) and Learning Nested Logit (L-NL) models. Using multiple publicly available datasets based on revealed and stated preferences, we show that our models outperform the traditional ones, both in terms of predictive performance and accuracy in parameter estimation. All source code of the models are shared to promote open science.
Within the United States, there have been evolving perceptions on the benefits and disadvantages of using peaked flight schedules. Arguments in favor of using peaked schedules have centered on a ...traditional assumption that consumers prefer itineraries with the shortest connection. However, prior work based on stated preference survey data has suggested that consumers avoid itineraries with the shortest possible (or minimum) connection times, and prefer those that add in an additional buffer of up to 15 min. In this study, we use a revealed preference dataset based on ticketing data from major U.S. carriers and find that, on average, consumers prefer itineraries that add in an additional buffer time of up to 25 min. Consumer preferences for buffer times beyond 25 min are less clear, with the exception of markets that are less than 600 miles apart, in which we see consumers are more sensitive to longer buffer times (and by extension, longer connection times). Our results can be used to help inform depeaking decisions within U.S. markets.
•We use a U.S. ticketing database to model consumers' connection time preferences.•We find that buffer times positively impact consumers' utility up to about 25 min.•Further, we find the effects of buffer time on consumers' utility beyond 25 min is more ambiguous.•Our results can be used to provide guidance for depeaking initiatives.