Hyatt Centric Fisherman's Wharf San Francisco, CA USA
Hyatt Centric Fisherman’s Wharf is centrally located near San Francisco landmarks including Pier 39, Ghirardelli Square, and the historic network of cable cars. Venture out to the Golden Gate Bridge and beyond from your convenient accommodations in Fisherman’s Wharf.
Adress: 555 North Point Street,
San Francisco, California, USA, 94133
Tel: +1 415 563 1234
Fax: +1 415 486 4444
Cut-off date to make a reservation at group rate is April 24, 2017. -NOW CLOSED
Use the link below to make room reservations at the conference group rate.
After the cut-off date, any rooms available will be at the regular Hyatt rates and must be booked through other channels.
Call For Presentations
You are invited to submit a proposal to present at the meeting. Come and share with us your ideas, thoughts, practical innovations, current trends, case studies, philosophies, and latest advances on the topics which are relevant to you.
Abstract Submission - please contact email@example.com
Final Presentation Due: May 10th
Details of the agenda will be updated once finalized.
May 16, 2017, 6:00 to 8:00 pm
May 17, 2017, 7:00 to 21:00 pm
Business Events - Business Casual
Welcome Reception - Business Casual
Social Events - Evening Casual
WeatherSan Francisco weather is mild. Plan for low 45F/8C and high 70F/21C with clouds and small chance of rain.
Situated on a peninsula separating San Francisco Bay from the Pacific Ocean, San Francisco is a uniquely picturesque city, whose scenic attractions include the largest cultivated urban park in the country, Golden Gate Park. Its notoriously steep streets, traversed by the famous cable cars, are home to a remarkably diverse ethnic population, and the city's reputation for tolerance and diversity is also evident in its history as a mecca for the gay community. Known for sophisticated cultural innovation and experimentation, San Francisco was the gathering place of the "beat" generation in the 1950s and a focal point of the 1960s counterculture, a hotbed of political protest and the birthplace of the "San Francisco Sound." Still known for its cultural attractions, today the Bay Area is also famous for its concentration of cutting-edge high-technology firms, which have drawn even more new residents to this populous region.
(Info from encyclopedia.com)
The United States requires visas or electronic authorization for travel from most countries. You can obtain more information at https://travel.state.gov/content/visas/en.html
If you need a visa, please check the wait time at your consulate and allow plenty of time. If you need supporting documentation for your application, please contact the conference chairs.
Aligning Aircraft Rotations with Crew Pairings: the Effect on Operational Stability
Eduardo de la Mora – AeroMéxico
Tomas Larsson – Jeppesen
In most cases the minimum turn time for an aircraft is shorter than the minimum connection time for crew, when crew need to change aircraft. When crew can follow aircraft, pairings will be both more productive and stable, given that the aircraft rotations hold through day of operation. In many cases tail assignment is done late, days before day of operation, and aircraft rotations are broken in order to meet maintenance requirements.
AeroMéxico operates a mix of regional, short-haul and long-haul operation. Especially the regional operation is very tight, and even small delays can create significant ripple effects. Those ripple effects can be contained much better if crew can follow aircraft. To improve punctuality, AeroMéxico starts the tail assignment process early, and use crew pairings as a side constraint to the tail assignment process. Tail assignments are then re-optimized throughout the month. This has resulted in improved stability on day of operation. In this presentation we will describe the process, and the impact on stability as observed from measurements on operational performance.
Codeshare Alliance Evaluation based on Network Optimization
Umut Besikci, Mauro Piacentini, Hunkar Toyoglu, Joakim Kalvenes – Sabre Airlines Solutions
Codeshare alliances are an important part of the network of many airlines and decisions related to codeshare partnerships are closely coupled with Revenue Management (RM) processes and decisions. The evaluation of these RM processes and decisions requires sophisticated tools that should account for codeshare alliances and general network features as well as demand dependency relationships. In this study, a codeshare alliance evaluation tool is proposed based on a network RM optimization formulation. Conceptually, an airline network consists of different itineraries that can be a combination of online and/or codeshare flights. The well-known Sales-Based Linear Program (SBLP) is used to calculate the maximum achievable revenue and catch spill-recapture effects between dependent alternatives under flight-capacity and demand restrictions. The model has been generalized to include codeshare services and different types of partnership agreements, such as bid-price exchange and free-sell with their different characteristics. In order to evaluate a specific codeshare alliance, two different scenarios are solved and compared. The first scenario corresponds to the original network optimization model where codeshare sales are allowed. This model is used to calculate the maximum achievable revenue over the network including the codeshare sales. In the second scenario, the network optimization model is modified so that the codeshare partner flights are removed from the network. In this scenario, it is possible to evaluate how the codeshare demand can be partially recaptured within the available online alternatives, as well as how the freed capacity will be used for alternative demand. Comparing the two scenarios produces important information and insights for correctly evaluating opportunity costs or revenue benefits, based on codeshare partnership activation or dismissal.
Airline CEO Challenge: Strategic Planning and Decision-making in a competitive environment.
Leandro Serino, Pablo Sirlin – Business Skills
We develop a business simulation model that replicates the operation of an airline. The model is designed to understand key aspects of strategic planning and decision-making in a competitive environment. Users are challenged to define a business strategy and make decisions in the areas of network planning, pricing, revenue management, ancillary revenue, product quality, fleet planning and asset financing. Taking into account information about customer preferences, market trends and a set of key performance indicators, which include the company’s credit ratings, users are expected to make evidence-based decisions and deal with the trade-offs commonly involved in business decisions. Trade-offs may be related to the product-quality mix and decisions about which technologies are worth implementing to improve the customer experience, turn-around times to improve passenger connectivity and/or the use of aircrafts, the pricing and marketing mix and whether to buy or lease the aircrafts, among others. In our three-period simulation model, the performance of the company will depend on the consistency of business decisions (over time and among departments of the company) and the ability to offer a differentiated product in a dynamic and competitive environment.
The opportunity of improving network planning by incorporating search data and geo-localization as an initial step before calculating a market size and allocating it to a departure airport
Faical Allou - Skyscanner Ltd
Data currently available for network planning is based on booking/ticket/boarding and most network planning algorithms are based on the following premises:
-Traffic from an airport is independent from the service offered (zero-sum game + “natural” growth)
-Traffic from an airport is not transferable to another airport outside city codes or manual groupings
We believe that:
-Including the shopping data, through the analysis of conversion rate can help estimate how much of the unconstrained potential can be converted to travel
-Users should be allocated to their city of residence/work and model their choice for departure airport depending on the service offer
Through IP addresses, albeit with flaws, the Internet has enabled localization of users in a more accurate way than the agency identifiers during the GDS era. In addition, modern data capabilities enable to store and process bigger datasets which made possible the use of shopping data (search).
Many air travelers are sensitive to service and would chose to not travel if the service is not what they expect service being a combination of price and schedule; this is seen with a significant difference in conversion rate. Air travelers are increasingly savvy and can travel by ground, or on two separate tickets to get a better price. Traditional measurements will allocate the “demand” to the wrong airport; and on the wrong O&D as shown below when isolating users from a specific city and breaking down their itineraries by origin airport of the last “ticket”.
Airline Market Segmentation
Himadri Mukherjee, Lakshmi Manasa Kasivajjula, Vineeth Patapati – Sabre Airline Solutions
The current airline industry is both competitive and dynamic. Maintaining profits in such an environment is quite a challenge without proper planning, marketing and operations for the airlines. Planning becomes more and more important with such dynamic system, where competitive and flight characteristics keep changing with technology enhancements. Thus, tools for planning and scheduling become integral part of these teams within airlines. Most of such tools and the analyses performed are designed to pick all the input parameters into the system at entity level. Entity is a group of OD pairs, usually grouped based on their geographical similarities.
Market (Origin-Destination Pair) segmentation or grouping is used in many facets of research and planning today by all the airline planning and scheduling teams. But, these market groups are more or less based on geographical characteristics, for example groups of OD pairs that originate in a particular region like US and terminate in another like Middle East. Various commercial tools used for planning and scheduling are usually designed to pick market groups that are geography based.
The tool used for the current research work is Sabre AirVision Profit Manager which provides flexible decision support system for evaluating the profitability of a proposed flight schedule using entity based input parameters. A calibration process generates input parameters into the PM at an entity level. Calibration is a seasonal and iterative process to estimate input parameters for connection formation, market share estimation and other key competitor parameters computation for schedule profitability estimation, usually computed at a market group level.
These market groups heavily depend on various other non-geographical characteristics of the airports, itineraries and schedules. Thus not considering them leads to heterogeneous market grouping. Proving the same, the current calibration process requires an additional effort during the final stages during which the analysts work towards generating individual market level parameters, for those markets which behave differently from what the entity group is expected to be like.
This research work demonstrates the higher efficiency of market grouping based on non-geographical market characteristics as compared to the typical market grouping methods. The presentation would cover the types of attributes selected and kind of clustering methodology that was adopted in detail. Comprehensive results, which display the clear cut difference among the groupings before and after will also be discussed during the session.
LARCH: A package for estimating multinomial, nested, and cross-nested logit models that account for semi-aggregate data
Laurie Garrow, Jeff Newman - Georgia Institute of Technology
Virginie Lurkin - École Polytechnique Fédérale de Lausanne
We present a summary of important computational issues and opportunities that arise from the use of semi-aggregate data (where the explanatory data for choice scenarios are not necessarily unique for each decision-maker) in discrete choice models. This data feature is commonly encountered with large transactional databases that have limited consumer information, such as itinerary choice modeling. We developed a software package called Larch, written in Python and C++, to take advantage of this kind of data to greatly speed the estimation of discrete choice model parameters. Benchmarking experiments against Stata (a commonly used commercial package) and Biogeme (a commonly used freeware package) based on an industry dataset for airline itinerary choice modeling applications shows that the size of the input estimation files are 50 to 100 times larger in Stata and Biogeme, respectively. Estimation times are also much faster in Larch; e.g., for a small itinerary choice problem, a multinomial logit model estimated in Larch converged in less than one second whereas the same model took almost 15 seconds in Stata and more than three minutes in Biogeme.
Estimation of Airline Itinerary Choice Models Using Disaggregate Ticket Data
Virginie Lurkin - École Polytechnique Fédérale de Lausanne
Laurie Garrow, Matt Higgins, Jeff Newman - Georgia Institute of Technology
Airline itinerary choice models support many multi-million dollar decisions, i.e., they are used to evaluate potential route schedules. Classic models suffer from major limitations, most notably they use average fare information but to not correct for price endogeneity. We use a novel database of airline tickets to estimate multinomial logit itinerary choice models using detailed fare data and compare these to classic itinerary choice models that use aggregate fare information but correct for price endogeneity. We extend the analysis to nested logit and ordered generalized extreme value models and find - much to our surprise – the competition structures captured in advanced discrete choice models have remained stable over the past 15 years. One of the key limitations with existing itinerary choice models is that they do not incorporate different choice set generation rules, e.g., they do not model how individuals filter or screen alternatives prior to making a decision. We explain how we designed an online experiment to model this filtering and screening behavior using Amazon Mechanical Turk (AMT).
Fleet Type Assignment and Robust Airline Scheduling with Controllable Cruise Times
Ozge Safak, M. Selim Akturk – Bilkent University
Sinan Gurel – Middle Eastern Technical University
In this study, we assign eet types to given ight paths during the booking period. We consider not only flight timing and passenger demand, as commonly done in the literature, but also operational expenses, such as fuel burn and carbon emission costs associated with cruise speed adjustment to ensure the passenger connections. We model the uncertainty in non-cruise times via a random variable arising in chance constraints to ensure the passenger connections. To achieve desirable connection probabilities, we simultaneously control cruise speed and allow minor adjustments on the flight departure times by redistributing the existing slack over vulnerable connections and removing excess slack from the remaining connections. Nonlinear fuel and carbon emission cost functions, chance constraints and binary aircraft assignment decisions make the problem significantly more difficult. To handle them, we use mixed-integer second order cone programming. We compare the performance of a schedule generated by the proposed model to the published schedule for a major U.S. airline. On the average, there exists a 20% overall operational cost saving compared to the published schedule. To solve the large scale problems in a reasonable time, we also develop a two-stage algorithm, which decomposes the problem into planning stages such as aircraft-path assignment and robust schedule generation, and then solves them sequentially. In further study, sequential planning approach of airline scheduling along with high stochasticity in flight operations naturally lead us to a multi-stage stochastic programming approach. We explicitly consider several potential demand and non-cruise time scenarios in making fleeting, routing and timing decisions so as to obtain improved solutions that would lead to substantial savings over the deterministic approach.
Integrating Fleet Assignment with Passenger Mix Models
Xiaodong Luo, Sergey Shebalov - Sabre Airline Solutions
We consider the integration of fleet assignment with various passenger mix models. The passenger mix models we consider can address network effect, demand recapture characteristics, the randomness in demand as well as the revenue management considerations with various level of granularity. The more demand information we consider, the more accurate the models are when compared to reality. However, the model accuracy comes with a price on the computational time, due to the increased difficulty of obtaining good quality solutions. To reduce solution time, we implemented some simple warm start techniques, heuristic fixing techniques as well as decomposition techniques. Using data from one of our client, we compare performance of some of these models, under various demand load, demand variability as well as equipment assignment cost scenarios. We also use Monte Carlo simulation to verify the revenue improvement and the cost savings of our models. Our study shows that it is possible to solve complex integrated fleet assignment model during flight scheduling phase. The combined increase in revenue and the reduction in fleet assignment cost can exceed tens of millions of dollars annually.
TNT: Total Market Traffic
Vivian Schobben - KLM Royal Dutch Airlines
MIDT data is frequently used in the analyses performed by our Network Planners. Unfortunately, these data represent only a part of the total market, since solely bookings of Travel Agents are in the GDSs’ records. The number of bookings made via direct channels (which are not in MIDT) have been increasing, thus resulting in the MIDT data becoming less and less representative of the total market. Our Network Planners have dealt with this issue by “manually” up-scaling the number of MIDT pax in Excel. However, they were very much aware of the shortcomings of their approach and the underlying assumptions. Therefore, the need for a more accurate estimate of the total market traffic between all origins and destinations arose. This is where we came in.
In this presentation, I will elaborate on the model we have developed and the results we achieved.