Artificial Intelligence and Machine Learning in the Travel Industry: Simplifying Complex Decision Making

Contents
Special issue on artificial intelligencemachine learning in travel
References
Price elasticity estimation for deep learning-based choice models: an application to air itinerary choices
Abstract
Introduction
Elasticity estimation from neural network-based choice models
Elasticity
Neural Networks
State-of-the-art deep network-based choice models
Automatic differentiation-based estimation
Airline itinerary choice dataset with price elasticity estimation and trip purpose segmentation
Price elasticity estimation from airline itinerary choices
Trip purpose segmentation: motivation
Trip purpose segmentation: dataset and experimental protocol
Analysis and business insights
Conclusions
References
An integrated reinforced learning and network competition analysis for calibrating airline itinerary choice models with constrained demand
Abstract
Introduction
Overall framework
Input data
Itinerary choice models
Classification of airport pairs
Explanatory variables
The airline network competition analysis (NCA) model
Parameters estimation using reinforced learning algorithm
Model application
Results
Model convergence
Error distribution
Model parameters
Summary and conclusions
References
Decoupling the individual effects of multiple marketing channels with state space models
Abstract
Introduction
Literature review
Model development
State space model
Baseline time series dynamics
Estimation
Application to simulated data
Optimum budget allocation
Conclusion
References
Competitive revenue management models with loyal and fully flexible customers
Abstract
Introduction
Motivation
Review of demand models that include competitive information
Modeling contexts
Class-based Bayesian learning demand model
Class-based forecasting with market information (market hybrid)
Bayesian dynamic linear model
Update equations
Availability covariates
Market hybrid examples
Market positions and optimal actions
Static and deterministic market scenario
Dominant market position
Medium market position
Class-free demand model with competitor information
Bayesian inference for the class-free demand model
Numerical study with airline data
Preliminary data analysis and reference competitor price
Qualitative analysis
Forecast accuracy analysis
Optimal pricing policy
Example: optimal pricing policy
Example: real airline data
Conclusion and future research
Acknowledgements
References
Demand estimation from sales transaction data: practical extensions
Abstract
Introduction
Practical limitations of existing models
Partial availability
Extending the attraction model
Data disaggregation
Comparison on a simulated example
Strong sell-down
Constrained parameter space
Non-homogeneous product set
Handling market share constraint and no-purchase option
Extended EM algorithm
Example: limited OA availability
Example: non-homogeneous product set
Constrained optimization
Solution using MM algorithm
Solution using Frank–Wolfe method
Example: non-homogeneous product set with constraint
Conclusions and future research
Acknowledgements
References
How recommender systems can transform airline offer construction and retailing
Abstract
Introduction
Recommender system use cases throughout the traveler journey
Next travel destination
FFP personalization
Search filtering and ranking
Upsell, cross-sell and third-party content
Advertised services
Airportflight experience
Toward a new distribution capability for the airline industry
Traditional distribution model
New distribution capability (NDC)
The offer management system (OMS)
The science of recommender systems
Introduction to recommender systems
Collaborative filtering recommender systems (CF)
Content-based filtering recommender system (CB)
Context-aware recommender system (CA)
Knowledge-aware recommender system (KA)
Session-based recommender system (SB)
Adapting recommender systems for offer construction and retailing
Next travel destination
FFP personalization
Search filtering & ranking
Upsell, cross-sell and third-party content
Advertised services
Airportflight experience
Conclusion and future research directions
References
A note on the advantage of context in Thompson sampling
Abstract
Introduction
Overview
Background
Thompson sampling for bandit problems
Thompson sampling for the multi-arm bandit
Thompson sampling with context
Pólya-gamma augmentation for binomial regression
Pólya-gamma augmentation for Thompson sampling
Simulation
Conclusion
References
Shelf placement optimization for air products
Abstract
Problem introduction
Problem description
Possible solutions
Clustering
Attribute based definitions
Cabin based approach
Solution details
Methods for defining optimal shelf structure
Optimization metric
Optimization technique
Auxiliary metrics
Future work
Experimentation on shelf definitions
Trip purpose segmentation
References
Applying reinforcement learning to estimating apartment reference rents
Abstract
Introduction
Related literature
Methodology
RL framework
Reference rent estimation algorithm
Performance measurement
Datasets
Empirical results
Conclusion
References
Machine learning approach to market behavior estimation with applications in revenue management
Abstract
Introduction
Future competitive schedule
Future market size
Future market share
Conclusion
References
Multi-layered market forecast framework for hotel revenue management by continuously learning market dynamics
Abstract
Introduction
Literature review
Research methodology
Data categories
Seasonality
Market segments
Group segment
Holidays and events
Span of data and sample size
Machine learning
Bootstrap aggregating
GradientBoost
Stacking
Feature engineering
Unconstrained and constrained demand
Modeling
Results and discussions
Forecast Interpretation
Conclusion
References
Artificial Intelligence in travel
Abstract
Overview
Technology hype cycle
Role of AI in travel
Revenue opportunity model–pattern recognition to recommend inventory control changes
Offer management
Demand forecasting
Seat pricing
Customer segmentation
Test and learn experimentation with the multi-armed bandit
Name recognition
Biometric boarding
Chatbots
Corporate travel and multi-lingual real-time translation
AI-enabled business
Overcoming complexity
Challenge of interpretability
Democratization of AI
Quantum computing and AI
Employees and AI
Conclusions
Acknowledgements
References
The key to leveraging AI at scale
Abstract
Introduction
Organizational barriers to scaling AI
Not having the right data architecture
Underestimating data science lifecycle
Unclear strategy to operationalize models
Insufficient business leader involvement
Insufficient executive sponsorship
Recommended best practices
Conclusion
The future of AI is the market
Abstract
New distribution capability
Customer value chain
We need to move on
Reinforcement learning
Distributed AI
Marketplace
Conclusion
Notes
References