Artificial Intelligence and Machine Learning in the Travel Industry: Simplifying Complex Decision Making Ben Vinod full chapter instant download

Artificial Intelligence and Machine
Learning in the Travel Industry:

Simplifying Complex Decision Making
Ben Vinod
Visit to download the full and correct content document:
https://ebookmass.com/product/artificial-intelligence-and-machine-learning-in-the-trav
el-industry-simplifying-complex-decision-making-ben-vinod/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Analytics Enabled Decision Making Vinod Sharma
https://ebookmass.com/product/analytics-enabled-decision-making-
vinod-sharma/
Risk Modeling: Practical Applications of Artificial

Intelligence, Machine Learning, and Deep Learning
Terisa Roberts
https://ebookmass.com/product/risk-modeling-practical-
applications-of-artificial-intelligence-machine-learning-and-
deep-learning-terisa-roberts/
Artificial Intelligence and Machine Learning for EDGE

Computing 1st Edition Rajiv Pandey
https://ebookmass.com/product/artificial-intelligence-and-
machine-learning-for-edge-computing-1st-edition-rajiv-pandey/
Sustainable Developments by Artificial Intelligence and

Machine Learning for Renewable Energies Krishna Kumar
https://ebookmass.com/product/sustainable-developments-by-
artificial-intelligence-and-machine-learning-for-renewable-
energies-krishna-kumar/
Visualization Techniques for Climate Change with
Machine Learning and Artificial Intelligence Arun Lal
Srivastav
https://ebookmass.com/product/visualization-techniques-for-
climate-change-with-machine-learning-and-artificial-intelligence-
arun-lal-srivastav/
Machine Learning for Decision Makers: Cognitive

Computing Fundamentals for Better Decision Making 2nd
Edition Patanjali Kashyap
https://ebookmass.com/product/machine-learning-for-decision-
makers-cognitive-computing-fundamentals-for-better-decision-
making-2nd-edition-patanjali-kashyap/
Machine Learning for Decision Makers: Cognitive

Computing Fundamentals for Better Decision Making 2nd
Edition Patanjali Kashyap
https://ebookmass.com/product/machine-learning-for-decision-
makers-cognitive-computing-fundamentals-for-better-decision-
making-2nd-edition-patanjali-kashyap-2/
Predicting Heart Failure : Invasive, Non-Invasive,

Machine Learning, and Artificial Intelligence Based
Methods 1st Edition Kishor Kumar Sadasivuni
https://ebookmass.com/product/predicting-heart-failure-invasive-
non-invasive-machine-learning-and-artificial-intelligence-based-
methods-1st-edition-kishor-kumar-sadasivuni/
Automation and Autonomy: Labour, Capital and Machines

in the Artificial Intelligence Industry 1st ed. 2021
Edition Steinhoff
https://ebookmass.com/product/automation-and-autonomy-labour-
capital-and-machines-in-the-artificial-intelligence-industry-1st-
ed-2021-edition-steinhoff/
Artificial Intelligence and
Machine Learning in the
Travel Industry
Simplifying Complex
Decision Making
Edited by
Ben Vinod
Artificial Intelligence and Machine Learning
in the Travel Industry
Ben Vinod
Editor
Artificial Intelligence and Machine

Learning in the Travel Industry
Simplifying Complex Decision Making
Previously published in Revenue and Pricing Management

Special Issue “Artificial Intelligence and Machine Learning
in the Travel Industry” Volume 20, Issue 3, March 2021
Editor
Ben Vinod
Grapevine, TX, USA
Spin-off from Journal: “Artificial Intelligence and Machine Learning in the Travel Industry” Volume 20, Issue 3, March 2021
ISBN 978-3-031-25455-0
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned,
specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter
developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific
statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of
publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for
any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional
affiliations.
This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Contents
Special issue on artificial intelligence/machine learning in travel ..................................................................................... 1

B. Vinod: Journal of Revenue and Pricing Management 2021, 2021: 20:211–212
(13, March 2021) https://doi.org/10.1057/s41272-021-00307-0
Price elasticity estimation for deep learning‑based choice models:
an application to air itinerary choices.................................................................................................................................. 3
Rodrigo Acuna‑Agost, Eoin Thomas and Alix Lhéritier: Journal of Revenue
and Pricing Management 2021, 2021: 20:213–226 (22, March 2021)
https://doi.org/10.1057/s41272-021-00308-z
An integrated reinforced learning and network competition analysis
for calibrating airline itinerary choice models with constrained demand ...................................................................... 17
Ahmed Abdelghany, Khaled Abdelghany and Ching‑Wen Huang: Journal
of Revenue and Pricing Management 2021, 2021: 20:227–247 (15, March 2021)
https://doi.org/10.1057/s41272-021-00309-y
Decoupling the individual effects of multiple marketing channels
with state space models ....................................................................................................................................................... 39
Melvin Woodley: Journal of Revenue and Pricing Management 2021, 2021:
20:248–255 (7, April 2021) https://doi.org/10.1057/s41272-021-00310-5
Competitive revenue management models with loyal and fully flexible customers ....................................................... 47
Ravi Kumar, Wei Wang, Ahmed Simrin, Sivarama Krishnan Arunachalam,
Bhaskara Rao Guntreddy and Darius Walczak: Journal of Revenue and Pricing
Management 2021, 2021: 20:256–275 (24, March 2021)
https://doi.org/10.1057/s41272-021-00311-4
Demand estimation from sales transaction data: practical extensions ........................................................................... 67
Norbert Remenyi and Xiaodong Luo: Journal of Revenue and Pricing Management 2021,
2021: 20:276–300 (22, March 2021) https://doi.org/10.1057/s41272-021-00312-3
How recommender systems can transform airline offer construction and retailing ..................................................... 93
Amine Dadoun, Michael Defoin‑Platel, Thomas Fiig, Corinne Landra and Raphaël Troncy:
Journal of Revenue and Pricing Management 2021, 2021: 20:301–315 (20, March 2021)
https://doi.org/10.1057/s41272-021-00313-2
A note on the advantage of context in Thompson sampling........................................................................................... 109
Michael Byrd and Ross Darrow: Journal of Revenue and Pricing Management 2021,
Shelf placement optimization for air products ................................................................................................................ 115
Tomasz Szymanski and Ross Darrow: Journal of Revenue and Pricing Management 2021,
Applying reinforcement learning to estimating apartment reference rents ................................................................. 123
Jian Wang, Murtaza Das and Stephen Tappert: Journal of Revenue and Pricing Management
2021, 2021: 20:330–343 (16, March 2021) https://doi.org/10.1057/s41272-021-00316-z
v
vi Contents
Machine learning approach to market behavior estimation with applications

in revenue management .................................................................................................................................................... 137
Nitin Gautam, Shriguru Nayak and Sergey Shebalov: Journal of Revenue
https://doi.org/10.1057/s41272-021-00317-y
Multi‑layered market forecast framework for hotel revenue management
by continuously learning market dynamics ..................................................................................................................... 145
Rimo Das, Harshinder Chadha and Somnath Banerjee: Journal of Revenue
https://doi.org/10.1057/s41272-021-00318-x
Artificial Intelligence in travel .......................................................................................................................................... 163
B. Vinod: Journal of Revenue and Pricing Management 2021, 2021: 20:368–375
(17, March 2021) https://doi.org/10.1057/s41272-021-00319-w
The key to leveraging AI at scale ...................................................................................................................................... 171
Deborah Leff and Kenneth T. K. Lim: Journal of Revenue and Pricing Management 2021,
The future of AI is the market .......................................................................................................................................... 177
Ross M. Darrow: Journal of Revenue and Pricing Management 2021, 2021: 20:381–386
(16, March 2021) https://doi.org/10.1057/s41272-021-00321-2
Journal of Revenue and Pricing Management (2021) 20:211–212
https://doi.org/10.1057/s41272-021-00307-0
EDITORIAL
Special issue on artificial intelligence/machine learning in travel

B. Vinod1
Published online: 13 March 2021

© The Author(s), under exclusive licence to Springer Nature Limited 2021
Over the past decade, Artificial Intelligence has proved efforts. Companies invest broadly across different platforms
invaluable in a range of industry verticals such as auto- such as Google or Facebook, and it is important to under-
motive and assembly, life sciences, retail, oil and gas, and stand the marginal impact of a specific marketing program
travel. The leading sectors adopting AI rapidly are Finan- or initiative on revenue. He uses a novel approach by casting
cial Services, Automotive and Assembly, High Tech and the well-known Koyck distributed lag model in state space
Telecommunications. Travel has been slow in adoption, but form to analyze the effectiveness of each marketing channel
the opportunity for generating incremental value for AI over and subsequent allocation of marketing budgets.
other analytics is extremely high (Chui et al. 2018). The paper by Ravi Kumar, Wei Wang, Ahmed Simrin,
In September 2019, Ian Yeoman and I discussed creat- Sivarama Krishnan Arunachalam and Bhaskar Rao Gun-
ing a special issue for the Journal of Revenue and Pricing treddy and Darius Walczak on competitive revenue manage-
Management on Artificial Intelligence in Travel. Information ment models is a collaboration between PROS and Etihad
from airlines and vendors on AI in travel has been sporadic, Airways. Their paper proposes a demand model that captures
usually discussed at industry conferences. Yet it was abun- realistic competitive dynamics by considering two types of
dantly clear to me based on my interactions with travel sup- customer behaviors: airline’s loyal customers who prefer to
pliers, software vendors, OTAs and GDSs that they were lev- buy from the airline even if their price is not the lowest in the
eraging core concepts in Artificial Intelligence and Machine market and fully flexible customers who buy the lowest fare
Learning to create new value propositions or improve on in the market. They develop a Bayesian machine learning-
existing applications related to travel. This was an opportu- based demand forecasting methodology for these models in
nity to showcase in a single issue the breadth and scope of both class-based and class-free settings that explicitly con-
what individuals in these organizations were focused on with siders competitive market information.
applications and business process. Norbert Remenyi and Xiaodong Luo from Sabre discuss
The research papers… practical limitations of the choice-based demand models
An excellent contribution from Rodrigo Acuna-Agost, found in the literature to estimate demand from sales trans-
Eoin Thomas and Alix Lh’eritier from Amadeus, who action data. They propose modifications and extensions
propose a new method to estimate price elasticity for deep under partial availability and extend the Expected Maximi-
learning-based choice models with an excellent set of refer- zation (EM) algorithm for nonhomogeneous product sets.
ences. The insights they provide are particularly relevant for The data preprocessing and solution techniques are useful
airline offers based on customer segment and context. for practitioners.
Ahmed Abdelghany and Ching-Wen Huang from Embry- The practice papers…
Riddle University and Khaled Abdelghany from Southern The paper on recommender systems by Amine Dadoun,
Methodist University propose a novel reinforcement learning Michael Defoin Platel, Thomas Fiig, Corinne Landra and
approach to calibrate itinerary choice models and measure Raphael Troncy from Amadeus highlight the central role of
schedule profitability. recommender systems to create personalized offers and its
Melvin Woodley from Sabre solves the attribution prob- growing importance with IATA’s New Distribution Capa-
lem of associating sales or revenue to individual marketing bility messaging standard. It is a well-researched paper and
truly relevant to the future of airline retailing.
Michael Byrd from Yum! and Ross Darrow from Charter
* B. Vinod and Go make the case for contextual bandits, a reinforce-
benvinod@yahoo.com ment learning technique for personalizing offers in retailing.
1
Southlake, USA They provide insights into the use of Thompson sampling, a
Chapter 1 was originally published as Vinod, B. Journal of Revenue and Pricing Management (2021) 20:211–212. https://doi.org/10.1057/s41272-021-
00307-0.
Reprinted from the journal 1

B. Vinod
popular exploration heuristic and how they can be deployed. Foremost on the minds of corporations as they leverage
They discuss the step improvement that can be achieved with AI for competitive advantage is how to scale AI across the
contextual bandits, despite greater computational complexity organization. Deborah Leff and Kenneth Lim from IBM
incurred when contextual features are included in the model. draw upon their extensive experience working with many
Tomasz Szymanski from Nordea Bank and Ross Darrow companies to provide insights into the various organizational
from Charter and Go discuss the important topic of shelf barriers to scale AI, the importance of executive sponsorship
placement on agency storefronts. While airlines focus on and recommend best practices. This paper is a “must read”
offer creation, the GDS desktop must display non-homoge- for anyone who is a practitioner of AI.
nous content that is addressed in this paper. A shelf product The futures article…
assortment method is proposed for categorizing airline offers Ross Darrow’s future’s article is thought provoking.
into utility levels, thus facilitating the itinerary selection pro- “The Future of AI is the Market” paints a picture of how the
cess for travelers. future travel distribution landscape will be influenced by
Jian Wang from Realpage outlines a practical application interactions in the marketplace and less on targeted one-off
of reinforcement learning used to determine reference rents solutions.
for apartments. He demonstrates how the new approach out- I would like to take this opportunity to thank all the anon-
performs the traditional rules-based approach. ymous referees I reached out to over the past few months
Shriguru Nayak, Nitin Gautam and Sergey Shebalov from to provide feedback on the submitted papers. This special
Sabre apply machine learning models to estimate market issue would not have been possible without your feedback
size and market share from competitive future schedules and and requests for revisions.
augmented data sources, a key component for developing
airline schedules. They also discuss how revenue manage-
ment practices can be improved with access to data from
network planning. Reference
The paper by Rimo Das, Harshinder Chaddha and Som-
nath Banerjee from LodgIQ focuses on forecasting market Chui, M., R. Chung, N. Henke, S. Malhotra, J. Manyika, M. Miremadi,
and P. Nel. 2018. Notes from the AI Frontier: Applications and
demand considering seasonality and market events. They Value of Deep Learning. McKinsey.com, April 2018.
examine a variety of machine learning techniques that the
data were calibrated upon and report on the accuracy of the Publisher’s Note Springer Nature remains neutral with regard to
forecasts. jurisdictional claims in published maps and institutional affiliations.
In January 2018, an AI initiative was established at Sabre
to identify industry-relevant problems suitable for AI-based
solutions, raise internal awareness and accelerate adoption. B. Vinod serves as Chief Scientist and Senior Vice President at
This initiative also led to the creation and distribution of an Sabre (2008–2020). Before rejoining Sabre in 2004, he was Vice
President at Sabre Airline Solutions, responsible for Pricing and Yield
internal AI newsletter, quarterly town halls to monitor pro-
Management.
gress and discuss use cases that I was responsible for. My
contribution to the special issue reflects this initiative and
steps taken to solve a range of problems in travel.
2 Reprinted from the journal

Journal of Revenue and Pricing Management (2021) 20:213–226
https://doi.org/10.1057/s41272-021-00308-z
RESEARCH ARTICLE
Price elasticity estimation for deep learning‑based choice models:

an application to air itinerary choices
Rodrigo Acuna‑Agost1 · Eoin Thomas1 · Alix Lhéritier1
Received: 10 May 2020 / Accepted: 4 September 2020 / Published online: 22 March 2021
© The Author(s), under exclusive licence to Springer Nature Limited 2021
Abstract
One of the most popular approaches to model choices in the airline industry is the multinomial logit (MNL) model and its
variations because it has key properties for businesses: acceptable accuracy and high interpretability. On the other hand,
recent research has proven the interest of considering choice models based on deep neural networks as these provide better
out-of-sample predictive power. However, these models typically lack direct business interpretability. One useful way to
get insights for consumer behavior is by estimating and studying the price elasticity in different choice situations. In this
research, we present a new methodology to estimate price elasticity from Deep Learning-based choice models. The approach
leverages the automatic differentiation capabilities of deep learning libraries. We test our approach on data extracted from
a global distribution system (GDS) on European market data. The results show clear differences in price elasticity between
leisure and business trips. Overall, the demand for trips is price elastic for leisure and inelastic for the business segment.
Moreover, the approach is flexible enough to study elasticity on different dimensions, showing that the demand for business
trips could become highly elastic in some contexts like departures during weekends, international destinations, or when the
reservation is done with enough anticipation. All these insights are of a particular interest for travel providers (e.g., airlines)
to better adapt their offer, not only to the segment but also to the context.
Keywords Price elasticity · Discrete choice modeling · Deep learning · Interpretability · Automatic differentiation · Travel
industry
Introduction performing tasks such as demand modeling and assortment

optimisation (Strauss et al. 2018). As more and more pur-
Discrete choice models describe the decision-making pro- chasing is made online, large volumes of purchasing choices
cess when choosing among a set of distinct alternatives by are tracked and stored, from which choice models can be
defining a probability distribution on them. These methods trained to select among competing offers. In these applica-
have been employed both to better understand the factors tions, the predictive power of the discrete choice model is
leading to decisions and to predict individual decisions. directly correlated to conversion rates and thus revenue. In
Historically, the use of discrete choice models has been such scenarios, predictive power is valued more highly than
extensively investigated in relation to travel choices and the interpretability of the model.
travel offer pricing (Garrow 2016). For example, analysis One of the oldest, and still widely used, approach is the
of discrete choice models learned from survey data has been multinomial logit (MNL) model that assumes that the choice
used to predict and inform new modes of transport such as probability is proportional to the utility of each alternative
new rail lines (McFadden 1974). Nowadays, choice models (Luce 1959; Block and Marschak 1960), which is equiva-
are integral parts of many revenue management systems, lently to satisfying Luce’s axiom also known as independ-
ence of irrelevant alternatives (IIA) (Duncan Luce 1977).
In McFadden (1973), the utility is defined as a function of
* Rodrigo Acuna‑Agost
rodrigo.acunaagost@amadeus.com observable attributes of the alternatives and some random
independent and identically distributed component, which
1
AI, Research and Technology Department, implies that a) all the decision makers behave similarly and
Amadeus S.A.S., 485 Route du Pin Montard, BP 69, b) the utility of the alternatives is independent.
06902 Sophia Antipolis Cedex, France
Chapter 2 was originally published as Acuna‑Agost, R., Thomas, E. & Lhéritier, A. Journal of Revenue and Pricing Management (2021) 20:213–226.
https://doi.org/10.1057/s41272-021-00308-z.

R. Acuna‑Agost et al.
Fig. 1 The proposed methodology to extract elasticity estimates from choice models. Note that the enrichment process is optional, and here it is
used to describe additional attributes of the data that are added for analysis, but not considered as features by the choice model
Natural evolutions of the MNL model deal with these (Lundberg and Lee 2017). A prediction can be explained by
limitations. For example, the mixed-MNL model allows assuming that each feature value of the instance is a player
grouping the decision makers by segments (Hayden in a game where the prediction is the payout. Shapley values
Boyd and Mellman 1980) and, similarly, the nested logit determine how to fairly distribute the payout among the fea-
model allows grouping alternatives by nests (McFadden tures. However, both LIME and Shapley values are relatively
1980). More recently, (Lhéritier et al. 2019) uses a machine expensive to compute, as they require a sampling of data
learning approach to allow an arbitrarily complex util- based on the decision to be explained.
ity function that can also depend on the decision maker’s Furthermore, the aim of this study is to extract mean-
attributes. ingful economic information, such as price elasticity1 from
Deep learning is a branch of machine learning that uses complex choice models, which are not provided by generic
artificial neural networks to model functions via an arbitrary model interpretation techniques. The proposed methodol-
number of composed transformations allowing to achieve ogy is summarized in Fig. 1. The first step is to estimate the
high performance in a large variety of tasks (see, e.g., Good- choice probabilities by the choice prediction model, then
fellow et al. 2016). Recently, some works used such approach elasticities are extracted for all the alternatives and choice
to build more flexible choice models. In Mottini and Acuna- situations. In order to give meaningful interpretation a filter-
Agost (2017), the authors propose a sequence transforma- ing step is needed to discard all non-relevant alternatives.
tion approach to define the choice probability allowing to Finally, the analysis is based on different aggregations using
condition on the decision maker’s attributes. More recently, a representative value (e.g., the median) per several dimen-
Lhéritier (2020) uses a deep learning approach to parameter- sions (e.g., customer segments and other features of interest
ize the flexible class of Pairwise Choice Markov Chains that depending on the application).
allows to escape traditional choice-theoretic assumptions The structure of the paper is as follows. We initially intro-
such as IIA, stochastic transitivity and regularity (Ragain duce the economic concept of elasticity and give an over-
and Ugander 2016). view of how it can be estimated for deep learning models.
For the application of air itinerary choice prediction, there We then present the motivating application of air itinerary
is a high interest in better understanding the choices of trave- choice modelling. This is followed by the numerical analysis
lers. This application of choice models can have important of elasticities for flight choices, and finally, we present future
impacts on revenue from using the most accurate models, research directions and conclusions.
but also by being able to get actionable insights that travel
providers like airlines or travel agencies could exploit to
improve their business metrics and offers. Elasticity estimation from neural
The high predictive performance of neural networks network‑based choice models
comes at the cost of a difficult interpretation of the models,
which has sparked research into complementary techniques In this section, we first introduce the economic concept of
to some shed light on how inputs attributes influence the elasticity and then we show how it can be computed on com-
outputs of the model. Various methods for performing fea- plex deep neural network-based choice models.
ture importance have been proposed (Molnar 2019), as well
as more specific methods to explain individual decisions.
Local interpretable model explanations (LIME) (Ribeiro
et al. 2016) can be used to explain a specific decision, by
building a linear model from samples close in the feature
space to the target sample. Shapley values have also been 1
Elasticity relates to the relative change of one variable (e.g.,
proposed as a method inspired by coalitional game theory demand) to the relative change in another variable (e.g., price).

Price elasticity estimation for deep learning-based choice models: an application to air…
Elasticity Compared to other industries (e.g., retail), the study

of price elasticity of demand in the air transportation has
Elasticity is an economic measure of how sensitive is a vari- remained relatively unexplored in the literature. Never-
able (e.g., demand) with respect to another one (e.g., price). theless we can highlight some of works: Jung and Fujii
More formally, the elasticity of x with respect to p at refer- (1976), Ghoshal (1981), Brons et al. (2002), Castelli et al.
ence values x0 and p0 is defined as (2003), Njegovan (2006), Richard (2009), Schiff and
Becken (2011), Granados et al. (2012), Granados et al.
𝜕x p0 (2012), and Morlotti et al. (2017).
𝜖px =
𝜕p x0
. (1)
Overall, these previous studies give similar results when
A particular case is when x corresponds to the quantity of it comes to order of magnitudes of price elasticities, vary-
demand of a given good with a price p. In that case 𝜖px rep- ing from inelastic values (− 0.3) to more elastic relations
resents the price elasticity of demand, a concept of special (− 2.0) depending on their context (different periods and/
importance in economics that will be leveraged in this work. or markets).
It should be noted that this value corresponds approxima- It should be noted that most previous works do not report
tively to the percentage change in the demand that is caused elasticities aggregated by relevant dimensions such as the
by a 1% change in the price of that good. trip characteristics (e.g., elasticity per day of the week).
Strictly speaking, elasticity can be either positive or It has been suggested that the price elasticity might vary
negative. However, in most cases elasticity is negative: according to the nature of the travel (Brons et al. 2002; Oum
when price increases, the demand decreases. However, et al. 1992; Morlotti et al. 2017) and the presence of substi-
economic theory provides two particular exceptions of tute modes (Brons et al. 2002). For example, Granados et al.
goods that defy common sense: Giffen and Veblen goods. (2012) reports a price elasticity of − 1.03, with difference
On the one hand, Giffen goods are associated to the case depending on the channel (online vs. offline) and different
of an inferior good (i.e., a good whose demand decreases market segments (business vs. leisure). The authors find
when consumer income rises) where a negative income that the values range from inelastic − 0.34 for business trip
effect induced by the price change is strong enough to booked offline to a more elastic value of − 1.56 for online
overcome a potential substitution effect (Spiegel 1994). channel for the leisure segment.
This was studied for the first time in 1815, when it was The contribution of this paper is the ability to provide this
reported that a rise in the price of bread corn, beyond a type of insights from any available dimensions for Neural
certain threshold, tended to increase the consumption of it, Network-based choice models.
as a consequence that people could not afford more expen-
sive substitutes (e.g., meat) (Heijman and Mouche 2011). Neural Networks
On the other hand, Veblen goods concerns normal goods
(i.e., demand increases when consumer income rises) that Deep feedforward networks are made of a series of layers
are particularly expensive and exclusive. On those prod- consisting of linear combinations of its inputs followed by
ucts, a higher price may make them desirable as a status a non-linear activation function. The coefficients and the
symbol, for example as observed in some luxury good and intercept of each linear combination are the free parameters
services (Veblen 1899). of the model, and their number define therefore its mod-
In terms of magnitude of elasticity, economists usually eling capacity. More complex architectures can be defined
look at their absolute value, classifying them into three by allowing other combinations of inputs or, for example,
groups: feedback loops as in recurrent neural networks. Thanks to
their versatility, deep networks can be used for a large vari-
• Elastic reaction, if 𝜖px is larger than 1. Examples: luxury ety of tasks, ranging from classical ones like classification or
items, vacations, high-end electronics, and generally regression to more complex ones like reinforcement learning
goods with many substitutes. (e.g., Bondoux et al. 2020). Some architectures are designed
• Inelastic reaction, if 𝜖px is smaller than 1. Examples: to reduce the dimension of the inputs in order to learn good
food, medicine, in general goods with vital importance representations of the features for the task under consid-
and few or no substitutes. eration, allowing to process complex inputs like images or
• Isoelastic reaction, if 𝜖px is equal to 1. text. The parameters of the linear combinations are typically
fit by minimizing a cost function on some training dataset
For an in-depth treatment, the reader is referred to, e.g., using some variant of the gradient descent algorithm, i.e.,
Dorman (2014) and Kolmar (2017). gradients of the cost function are taken with respect to the
parameters and these are moved in the negative gradient

direction using some step size. For an in-depth treatment

of deep neural networks, see e.g., Goodfellow et al. (2016).
State‑of‑the‑art deep network‑based choice models
Alternatives and individuals making choices can be

described by a set of features that can be then used to con-
dition the choice probability given by a choice model. A
linear-in-parameters MNL assumes that the representative
utility of each alternative is given by a linear combination of
its features (McFadden 1973). These simple MNL models
can be represented by a shallow neural network that uses
only one linear combination of the attributes of each alter-
native and a softmax function applied to each linear com-
bination. More sophisticated deep neural network-based
models have been proposed in the literature. In Hruschka
et al. (2001), a multi-layer extension has been proposed,
allowing to learn non-linear MNL models. In Mottini and
Acuna-Agost (2017), the authors propose a recurrent neural
network architecture with an attention mechanism that learns
to point, within a sequence of alternatives, to the chosen one.
PCMC-Net (Lhéritier 2020) is a deep network that param- Fig. 2 Computational graph obtained from PyTorch for a linear-in-
eterizes Pairwise Choice Markov Chains from the alterna- parameters MNL for three alternatives with two features represented
tives’ and the individuals’ features by combining a series in a 3 × 2 matrix x. The column vector w corresponds to the parame-
of modules performing representation learning, pairwise ters of the model and the column vector y represents the actual choice
with a 1 in the corresponding position and 0 elsewhere. Automatic
combination, standard feedforward processing and linear differentiation allows to obtain derivatives of the cost function c with
system solving to finally obtain the choice distribution. respect to w in order to fit the model and derivatives of the probability
PCMC-Net exhibits excellent predictive performance on a vector q with respect to inputs x in order to compute elasticities. qi
complex airline itinerary choice dataset where the alterna- denotes ith component of the vector q
tives are strongly dependent on an individual-specific query
and some features, like price, can vary with time. In order to basic differentiation rules and the chain rule to propagate the
understand which kind of behavioral properties are captured applied values through the graph (see Fig. 2 for an example).
by these complex deep networks, elasticities are of particu- It is different from numerical and symbolic differentiation,
lar interest. The probabilities that are provided by a choice and is both efficient (linear in the cost of computing the
model can be interpreted as market shares at equilibrium and value that is being differentiated) and numerically stable.
therefore can be used as demand quantities in Eq. 1. See, e.g., Baydin et al. (2017) for an in-depth treatment.
The derivative of the choice probability with respect to
Automatic differentiation‑based estimation some input usually depends on the point x whose coordi-
nates are the alternatives’ and the individual’s features of a
When using neural network-based choice models, prob- given choice situation. In order to interpret the predictions
abilities can be easily differentiated with respect to any of given by a model on some given dataset using elasticities,
the inputs using the automatic differentiation mechanism of derivatives can be computed on it and summarized by tak-
modern neural network libraries [e.g., PyTorch, see Paszke ing e.g., the median, as shown in the numerical results of
et al. (2019) or Tensorflow, see Abadi et al. (2016)]. An this paper.
automatic differentiation system converts a program specify-
ing operations (e.g., a neural network) into a computational
graph that represents it in terms of a composition of primi-
tive operations that have specified routines for computing
derivatives. Then, the derivative of some node (e.g., cor-
responding to the cost function or the probability given by a
neural network) with respect to some leaf node of the graph
(e.g., corresponding to parameters or inputs of the neural
network) on some given point x can be obtained by applying

Fig. 3 The methodology applied

to air itinerary choice. Price
elasticities are extracted from
the PCMC-Net choice model.
For analysis, only the alterna-
tive with the largest probability
of selection is selected in each
choice session. Trip purpose
segmentation is applied based
on a training set consisting of
bookings
Airline itinerary choice dataset with price Price elasticity estimation from airline itinerary
elasticity estimation and trip purpose choices
segmentation
A passenger name record (PNR) contains relevant data
Nowadays travelers have higher expectations and choice regarding travel bookings, such as flight information of
than in the past. This is mostly driven by the experience each segment of a journey and information about the indi-
they already have in other industries, in particular online vidual, as well as information about ancillary services and
retail. Some examples of the new standards are: relevant special service requests. In order to obtain a full choice set,
and timely recommendations, fully customized products
and services, transparent pricing, modern search, shopping
cart functionalities on different channels (e.g., mobile and Table 1 Features of the airline itinerary choice dataset
desktop), and high flexibility (e.g., be able to cancel sub- Type Feature Range/cardinality
scriptions at any time or to send back products and being
fully reimbursed). Many, if not all, of these improvements Individual Cat.Origin/destination 97
observed in retail have been boosted by leveraging data and Search office 11
newer algorithms thanks to machine learning, and naturally Num. Departure weekday [0, 6]
the travel industry is following this trend too. Stay Saturday [0, 1]
Figure 3 summarizes the application of the proposed Continental trip [0, 1]
methodology for getting elasticities and business insights Domestic trip [0, 1]
from air itinerary choice models. The remainder of this sec- Days to departure [0, 343]
tion presents the methodology to obtain this enriched dataset Alternative Cat. Airline (of first flight) 63
of chosen alternatives with elasticity and trip purpose esti- Num. Price [77.15, 16,781.50]
mates. The aggregations and business insights are presented Stay duration (min) [121, 434,000]
in the results section of this paper. Trip duration (min) [105, 4314]
Number connections [2, 6]
Number airlines [1, 4]
Outbound departure time [0, 84,000]
(in s)
Outbound arrival time (in s) [0, 84,000]

Fig. 4 Performance of different

heuristics and choice models on
the airline itinerary choice prob-
lem (Lhéritier 2020; Lhéritier
et al. 2019; Mottini and Acuna-
Agost 2017)
data from PNRs are matched with search log activity, which interested in price elasticity, that is, the elasticity of the
shows all available options presented to the traveller prior demand, estimated by the probability of being chosen, with
to booking. respect to the price. Moreover, we consider only one alterna-
In this experiment, the dataset from Mottini and Acuna- tive per choice situation. For each choice situation, we take
Agost (2017) consisting of flight bookings sessions on a set the alternative with the largest probability of selection as
of European origins and destinations is used. Each choice estimated by the PCMC-Net model. It should be noted that
session contains up to 50 different proposed itineraries, this alternative does not necessarily match the alternative
one of which has been booked by the customer. There are that was chosen by the consumer. This decision was taken as
815,559 distinct alternatives among which 84% are single- our main goal is to give explainability to the neural network-
tons and 99% are observed at most seven times. In total, based model, rather than understanding individual choices.
there are 33,951 choice sessions of which 27160 were used
for training and 6791 for testing. The dataset has a total of Trip purpose segmentation: motivation
13 features, both numerical and categorical, corresponding
to individuals and alternatives, as shown in Table 1. The first step to better modeling the traveller decision-
Choice models are important in helping to select, high- making process is to understand the reason why the traveler
light and rank different offers. Several methods have previ- would like to travel, which we will refer to hereafter as the
ously been suggested, the performance of which is shown trip purpose. If travel providers could get this information
in Fig. 4 on a common training and test dataset. The metric accurately at shopping time (i.e., before the booking), they
used here is the Top 1 accuracy, which measures the per- could greatly improve the shopping experience: offer the
centage of sessions for which the most probable alternative best product, at the best price, at the best moment, to the
identified by the model is indeed chosen by the user. Each targeted customer.
choice set contains 50 alternatives, thus a uniform sampling Business trips are driven by convenience and usually sub-
method achieves 2% Top 1 accuracy. Simple heuristics such ject to companies’ travel policies (i.e., the passenger do not
as selecting the cheapest offer can be used to give a baseline pay for this trip, but her company). It is also the case that
performance for the problem. As can be seen, non-linear sometimes the passenger has a less active role in the deci-
methods such as Deep Pointer networks, Latent Class MNL sion of the trip as the task is delegated to travel arrangers
and Random Forests give better performance than the linear such as travel agencies or assistants. The authors in Teichert
MNL model. However, PCMC-net results in the best Top 1 et al. (2008) confirm that people traveling for business have a
accuracy of all the methods tested. For more details on each strong correlation with these attributes: efficiency, punctual-
method, the reader is referred to Lhéritier (2020), Lhéritier ity, and flexibility.
et al. (2019) and Mottini and Acuna-Agost (2017). On the other hand, leisure trips are driven mostly by
Elasticity can be calculated for any reference value and price (Teichert et al. 2008). Although in some cases, the
for any alternative in the choice set. We are particularly passengers are sufficiently wealthy that they may give more

importance to comfort and efficiency. In practice, many trips are labeled as business travel. From this set of 400,000 book-
are not easily classified on exactly one of these two seg- ings, 40,000 random samples are held out as the test set for
ments as they can be both at the same time: bleisure trips the classification task.
(i.e., extending a business trip for leisure activities) (Vivion For each booking there are 48 features available corre-
2016). sponding to all non-sensitive attributes of the trip. These
As an important question for the industry, it is not surpris- relate to the origin, destination, route, carrier, various
ing that the problem has been addressed previously (Teichert aspects linked to the time and duration of travel along with
et al. 2008; Chatterjee et al. 2020; Tahanisaz and shokuhyar aspects of the booking such as the number of passengers
2020; Martinez-Garcia and Royo-Vela 2010; Jin-Long 2017; in the booking, the number of days prior to boarding that
Dresner 2006; Vinod 2008). Most of the previous work is the trip was booked, etc. A comparison of different models
based on stated preference surveys (Dresner 2006; Martinez- based on various feature subsets is provided in Appendix A.
Garcia and Royo-Vela 2010; Tahanisaz and shokuhyar 2020) In order to apply the segmentation of bookings to the
i.e., asking current or potential travelers about their prefer- choice dataset, only features common to both the bookings
ences and the reason of the trips. This kind of approach and choice datasets are selected. These are the origin and
brings a lot of flexibility in terms of the type of questions destination, international/domestic, stay duration, days to
as for example the analysts could even ask about hypotheti- departure, the day of the week for the booking, outbound
cal scenarios. It is well known that stated preference data flight and return flight and a stay Saturday feature.
present a series of inconveniences (Abdullah et al. 2011), A gradient boosting machine model (Friedman 2001)
for example, their incapacity to capture accurately all the is used as a classifier, with grid search and early stopping
market and personal limitations that occurs in the real world. selecting a maximum depth of 6 for 46 trees. On the hold-out
Another limitation of these works based on surveys is that test set, the model obtains 84.20% accuracy.2 Feature impor-
the conclusions are drawn based on relatively small amount tance analysis suggests that the stay duration, destination and
of data [around 3000 in Dresner (2006), 300 in Tahanisaz origin airports, stay Saturday, return day of week and days
and shokuhyar (2020), 808 in Martinez-Garcia and Royo- to departure are the most important features, respectively.
Vela (2010), and 5800 in Teichert et al. (2008)]. Note that this segmentation is not used as an input to the
Therefore, to adequately discuss the price elasticities choice model, but only to segment the dataset for analysis
obtained by the choice model, these should be done in the purposes (see Fig. 3).
context of business and leisure trip independently. However,
this information is not available from the choice dataset, and
as such must be inferred. Analysis and business insights
Trip purpose segmentation: dataset In this section, we analyze and discuss the elasticities
and experimental protocol obtained by the approach presented previously using differ-
ent aggregations.
In this section, we present an analysis of business vs leisure In order to aggregate the estimated elasticities, we use the
segmentation performed over a set of labeled bookings from median since it is a measure of central tendency robust with
a larger set of unlabeled data. The bookings correspond to respect to outliers and skewed data (see Fig. 5). Another
indirect bookings made by customers at traditional travel important element to remark is that the analyses are based on
agencies, online travel agencies and travel management com- the additive inverse of elasticity because the price elasticities
panies (among others) which are then processed by a GDS. are usually negative. This transformation helps to construct
We can consider the dataset as being partially labeled, as readable charts following the convention of economists that
most offices are identified as belonging to particular market are interested on the absolute values (magnitudes) instead
segments which deal almost exclusively with either business of the real number.
travel or leisure travel. Overall elasticity The median value for whole data is
The dataset used in this trip purpose model is a sam- − 1.73 which can be classified as elastic. Note that this value
ple of indirect bookings from the European market for the is in the same order of magnitude to the values published in
full year 2019. A balanced dataset is obtained by sampling previous works.
200,000 bookings made from offices tagged as “Retail- small Elasticity by trip purpose In order to understand better
medium enterprises” which are labeled as leisure travel, this value, we analyze the two main customer segments in
and sampling another 200,000 bookings made from offices
tagged as “Global Travel Management Companies” which
2
Accuracy: ratio of number of correct predictions to the total num-
ber of input samples.

Fig. 5 Distribution of elasticity values for both segments: Leisure additive inverse of the elasticity that is usually negative in its origi-
(left) and Business (right). The distributions are not symmetric, and nal form. It should be also noted the presence of few negative values
in both segments there is a peak of observations near to zero. Note (a) correspond to rare observations. These choices could be explained
we analyze one alternative per choice situation, the one with the high- by Veblen or Giffen behaviors of some consumers in certain circum-
est probability, (b) all charts (and this one in particular) present the stances
sensitivity, while in business trips we see a significant differ-

ence between working days and weekends. These results are
consistent with previous research reporting − 0.64 for work-
ing days and − 1.05 for weekends (see Morlotti et al. 2017).
Elasticity by domestic/international trip Figure 8 shows
the median elasticity on both segments split by domestic
Fig. 6 Overall median negative elasticity for both segments: Business
and international trips. Similarly to the previous chart, lei-
and Leisure
sure trips exhibit less variance in price elasticity, while,
for business trips, there is a significant difference between
the travel industry. Figure 6 shows the overall median elas- domestic and international trips. This could be explained by
ticity for leisure and business. As expected the median price travel policies in place in most companies, where it is com-
elasticity for leisure trips (− 6.71) is significantly greater mon to put price restrictions as a function of the distance
than the one concerning business trips (− 0.85). Indeed, the of the flights. Note that this particular dataset is related to
price elasticity for leisure trips is considered to be elastic, European markets, where international trips tend to be more
while the one for business is inelastic. The interpretation expensive than domestic trips.
of this value is that if the price is changed 1%, we should Elasticity by days to departure Another interesting analy-
expect a change in the probability of being chosen of approx- sis is presented in Fig. 9. The chart shows the median elastic-
imately 6.71% for a consumer looking for leisure trips (and ity as a function of the trip advance purchase, also known
0.85% for business trip, respectively). as days to departure (DTD). It is interesting to remark, as
Elasticity by day of the week Figure 7 shows median expected, that the price elasticity increases if the consumer
elasticities on both segments per day of the week (DOW). has more time to decide before the departure of the trip.
The results confirm the difference between both segments. Another interesting insight is that business and leisure trips
More interesting are the values on different DOW. Consum- seem to converge to similar values for large DTD. On the
ers related to leisure trips seem to have a more regular price other hand, it is clear that for both segments, trips happening

Fig. 7 Median elasticity (neg) for both segments (Business and Leisure) on different departure days of the week
in the near future (short-term) show very low price elas- Saturday and in blue for when the traveler did not stay at
ticities, probably explained as trips booked for closer dates destination a full Saturday. Note that the blue curve rises
are particularly related to urgent matters, in those cases the in a linear fashion, indicating that stay duration is directly
price is less relevant than other aspects like the schedule proportional to price elasticity for trips which return dur-
or the total trip duration. This is consistent with previ- ing the same week as departure (this also includes trips
ous research where elasticity was reported in the interval leaving Sunday and returning prior to the following Sat-
𝜖 ∈ {− 2.0, − 0.5} for DTD ∈ {2, 21} days (see Morlotti et al. urday). For trips including Saturday stays (in orange),
2017). Our results extend the previous results showing that all trips are highly price elastic, but it does appear that
the absolute value continues to increase to larger values until for stay durations between 1 and 3 nights are less price
100 days approximately. The chart also shows an increase on elastic. These short trips always contain a Saturday night
the dispersion of values for larger DTD, which is explained stay, and therefore often correspond to weekend geta-
mainly by the number of observations used to calculate the ways and possibly city breaks. Such trips are often to
median values (represented by the darkness of the line). geographically closer destinations, which are associated
Elasticity by stay duration An important criterion with cheaper prices overall, and thus price may be a less
in price elasticity is the stay duration, especially when important factor than for longer duration trips which can
weekend stays are factored into the analysis. In Fig. 10, be associated with a higher overall budget. Furthermore,
the price elasticity is shown as a function of stay dura- for such short trips, other factors such as time of arrival
tion in orange for trips where the traveler stayed an entire and departure may be more important to the travelers in

can be inelastic or not. Some cases where we observed

elastic reaction on business trips was on international
trips, departures during weekends, and reservations done
with enough anticipation (i.e., more than 30 days). This
could be explained by travel policies in place on most
companies. All these insights could be of the particular
interest on pricing strategies and other initiatives done by
travel providers to better adapt their offer to the market
conditions.
We suggest the following ideas that can inform future
research: (a) use elasticity as a new manner to segment pas-
sengers and (b) estimate other economic metrics of inter-
est, in particular the willingness to pay (WTP) for certain
features.
With respect to segmentation, the most common way to
group passengers in the travel industry has been to separate
business and leisure trips. These two segments allow dif-
ferentiation of products, but assume a clear difference in the
price elasticities and willingness to pay of the segments. As
concluded in this work, the business segment may become
highly elastic in some context. In that regard, we believe a
new way to segment customers could be leveraged from this
work, thus allowing segmentation of travel requests depend-
ing on the price elasticity that can be estimated in real time.
Fig. 8 Median elasticity (neg) for both segments (Business and Lei-
This would represent a new opportunity for dynamic pricing/
sure) on domestic and international trips packaging.
With regard to economic metrics, we have presented the
calculation of price elasticity in this work. Nevertheless,
order to maximise their time at destination, thus reducing there are other economic metrics that are important in the
the importance of price. industry. One of the them is the willingness to pay, usually
defined as the maximum amount of money a consumer is
willing to hand over to buy a product or service (Lu and
Conclusions Shon 2012; Carlos Martín et al. 2008; Chang and Sun 2012;
Carlsson 1999; Tsamboulas and Nikoleris 2008; Merkert and
In this paper, we have proposed a new methodology to Beck 2017). It should be noted that a way to estimate WTP
extract price elasticity from deep learning-based choice is by looking at the derivative of a feature with respect to the
models. The approach leverages the efficiency of auto- price, something that could be explored using the approach
matic differentiation capabilities of deep learning libraries. presented in this paper.
With this capability, we were able to estimate price elastic-
ities on all the data points (choice set). As a consequence,
the approach was flexible enough to permit deeper analysis Appendix: A trip purpose segmentation
in any dimension available in the data. With this in mind,
we focused on understanding price elasticity on the classi- To better understand the value of the business vs leisure
cal segments in the industry (business and leisure) and the segmentation proposed in this study, we showcase and com-
features/context that may affect their elasticity. pare three variations of the model. In the first approach, all
Regarding numerical experimentation, the elastici- features available in the dataset are used to generate the best
ties extracted from PCMC-Net suggest that the demand performing model and provide some analysis to understand
for air travel for the business and leisure segments dif- what characteristics of travel are important to differentiate
fers significantly. While business trips are price inelastic, different types of travel. In order to better understand the
leisure trips are estimated to be highly price elastic. The relationship between different features of travel, we also
presented methodology allowed to explore the elasticity showcase a simplified model based on a decision tree, with
on other dimensions of interest as well. For example, we minimal features and complexity, which can be fully inter-
found that, depending on different variables, business trips preted. Finally, we will also construct a model which can

Fig. 9 Median elasticity (neg) for both segments: leisure (left) and business (right) as a function of different advance purchase days (a.k.a. days
to departure)
Fig. 10 Median elasticity (neg)

as a function of stay duration,
for trips containing a Saturday
stay or not
be applied to a choice dataset used in the main article. This Gradient boosting models have proven particularly adept
choice data profile segmentation model is effectively a com- at classification, here the H2O.ai library is used to train
promise in performance, due to only some features overlap- the models and determine the feature importance (Candel
ping between the bookings dataset and the choice dataset. and Malohlava 2020). The training phase uses a hold-out

Table 2 Trip purpose experiments and results on hold-out test set

Model Model type No. of features Test
accuracy
(%)
Full feature set H2O GBM 48 86.5

Choice feature set H2O GBM 9 84.2
Reduced feature set Scikit learn DT 9 80.8
Note that the H2O package was used to train the gradient boosting
machines in order to obtain best performance, whereas Scikit-learn
was used to train a decision tree that could be visualized and inter-
preted
validation set to estimate the depth of the individual trees via

grid search and the number of trees via a stopping criterion.
In contrast, the interpretable model is trained with scikit
learn (Pedregosa et al. 2011) and consists of a single deci-
sion tree which was manually tuned to give an acceptable
tradeoff between human interpretability and performance
on the test set.
For the interpretable model features, these are selected
to represent attributes that are not overly dependent on the
market, e.g., carrier and destination country, but rather
those that are representative of general travel aspects, such
as stay duration and number of passengers in a booking.
Specifically, the features are limited to the number of pas-
sengers, distance of trip, international, stay duration, days
to departure, day of week of outbound and return flights as
well as booking date, and a feature indicating whether the
travelers stayed Saturday night during their travel.
The classification accuracy for the hold-out test set is
given in Table 2. The full feature set results are given in
Fig. 11 Interpretable decision tree for business vs leisure prediction.
order to set an upper bound estimate of the performance Nodes that are split show the feature and threshold used for the split.
given no constraints for the classification task. The accu- All nodes show the gini value, where lower values indicate purer
racy of 86.5% suggests that generally the trip purpose can nodes, the total number of training samples at the node, the number
of training samples from each class at the node [business, leisure],
be determined on this dataset, but there are bookings that
and the resulting class label. The color is based on the class label, and
are misclassified. The features ranked highest in terms of the intensity is a function of gini value. (Color figure online)
feature importance for this model are the carrier, destina-
tion country, whether the passengers stayed Saturday night
at destination, the number of passengers in the booking and depth of any branch, or by only allowing splits when suffi-
the advance purchase time prior to the flight. cient data are present before or after the split (often referred
In comparison, despite only accessing nine of the fea- to as minimum samples in split or minimum samples in leaf,
tures available, the choice feature set obtains 84.2%. Fea- respectively). Testing over different choices of parameters
ture importance analysis suggests that the stay duration, resulted in the choice of restricting the model complex-
destination and origin airports, stay Saturday, return day of ity using minimum samples before a split. This resulted
week and days to departure are the most important features, in the best accuracy on the hold-out test set for any model
respectively. This model is applied to the flight choice data- with under 10 splits, and also produced a model for which
set in order to provide additional context to the analysis in increasing complexity did not lead to large improvements in
the main article. accuracy over the test set, relative to other choices available
For the interpretable model, we use a single decision tree to restrict complexity.
with limited complexity. The complexity of the tree is con- This interpretable tree results in 80.8% accuracy on the
strained during training by limiting the number of splits per- same hold-out test set as used in the previous experiment.
formed in the tree. This can be done by setting the maximum

Another random document with
no related content on Scribd:
Tampico de Tamaulipas, Memoria sobre la evacuacion militar del
Puerto. [San Luis Potosí, 1848.]
Tápia (Andrés de), Relacion sobre la conquista de Mexico. In
Icazbalceta, Col. Doc., tom. ii.
Tápia (Eugenio de), Febrero Novísimamente redactado. Madrid,
1845-6. 8 vols.
Tarayre (E. Guillemin), Exploration Minéralogique des Régions
Mexicaines. Paris, 1869.
Tastera, et al., Carta al Emperador Don Cárlos. In Cartas de
Indias.
Taylor (Fitch W.), The Broad Pennant. New York, 1848.
Taylor (Fitch W.), The Flag Ship. New York, 1840. 2 vols.
Taylor (Isaac), The Mine. Philadelphia, 1841.
Taylor (Zachary), A brief review of the career, character, and
campaigns of. Wash, n.d.; A review of the life, character, and
political opinions of. Boston, 1848; A sketch of the Life and
character of. By the one-legged Sergeant. Boston, 1847; A
sketch of the Life and Public Services of. Wash. [1848];
Biography, A Collection; General Taylor’s moral, intellectual, and
professional character. Wash., n.d.; Letters of, etc. N. Y., 1848;
Life and Public Services of. By an officer of the U. S. A. N. Y.,
1846; N. Y., 1850; Life and Public Services of, also life and
services of Millard Fillmore. Hartford, 1848; Life, Battles, and
Despatches. Phil., 1847; The Life and Public Services of. Phil.,
n.d.; The Life of. N. Y., 1847; The Life of, and a history of the
War in Mexico. N. Y. 1847.
Taylor and Fillmore, Sketch of the lives of. Boston, n.d.
Taylor and His Campaigns, A Biography. Philadelphia, 1848.
Taylor and His Generals. Philadelphia, 1847.
Taylor and His Staff. Philadelphia, 1848.
Tehuantepec, An Account of the Isthmus of. London, 1846;
Análisis del Dictámen de la Comision sobre el negocio de. Mex.,
1852; Camino Carretero, Camino de Fierro y Canal por el istmo
de. Mex., 1870; Cuestion de Nueva-York, 1852; Decreto del
Supremo Gobierno para la apertura del Istmo. Mex., 1853;
Dictámen de la Comision Especial, con motivo del privilegio
concedido á José Garay. Mex., 1851; Dictámen de la Mayoría
de las Comisiones de Industria. Mex., 1852; Dictámen de la
Mayoría de la Comision especial de la Cámara de Diputados.
Mex., 1851; Discurso pronunciado por el Señor Fuente, 29 de
Noviembre. Mex., 1852; Documentos relativos á la apertura de
una via. Mex., 1852; Memoria instructiva de los derechos y
justas causas que tiene el Gobierno. Mex., 1852; Proyecto de
colonizar el istmo. [Mex., 1823]; Reglamento para el
Establecimiento de las Colonias Militares del Istmo. Mex. 1851;
Reglamento para la Comunicacion por la via Inter-Oceánica.
Mex., 1857; Right of way. n.pl. [1853]; Vindicacion de la
conducta política de los. Oaxaca, 1847.
Tehuantepec Canal, A Collection.
Tehuantepec Canal, A memorial setting forth the rights and just
reasons for not recognizing the validity of the privilege granted
to D. José Garay. New York, 1852.
Tehuantepec Canal, A Review of the Tehuantepec Controversy.
Georgetown, 1853.
Tehuantepec Railway, Its location, features, and advantages under
the La Sere Grant of 1869. New York, 1869.
Tepic, Manifiesto que el Ayuntamiento. Guadalajara, 1845; Santa
Cruz que se venera en el convento de Religiosos Franciscanos
de la ciudad. Guadalajara, 1853.
Ternaux-Compans (Henri), Recueil de Documents et Mémoires
originaux sur l’Histoire des Possessions Espagnoles dans
l’Amérique. Paris, 1840.
Ternaux-Compans (Henri), Voyages, Relations et Mémoires
originaux pour servir á l’histoire de la Découverte de l’Amérique.
Paris, 1837-41. 2 series, 10 and 8 vols.
Testimonio á la letra del Espediente mandado formar por el
Arzobispado de Cesarea. Mexico, 1848.
Testimonio de Hidalguía de Cortés. In Col. Doc. Inéd., tom. iv.
Testory (Abate), El Imperio y el Clero Mejicano. Mexico, 1865.
Tezozomoc, Recopilacion de tradiciones. MS.
Tezozomoc (Alvaro), Histoire du Mexique. Paris, 1853. 2 vols.
Thiers (A.), Historia del Consulado y del Imperio de Napoleon.
Mexico, 1845.
Thomas (Lewis F.), Cortez, the Conqueror. Washington, 1857.
Thompson (John Lewis), History of the wars of the United States.
Philadelphia, 1860. 2 vols.
Thompson (Waddy), Recollections of Mexico. New York, etc.,
1847.
Thorpe (T. B.), Our Army at Monterey. Philadelphia, 1847.
Thorpe (T. B.), Our Army on the Rio Grande. Philadelphia, 1846.
Thümmel (A. R.), Mexiko und die Mexikaner. Erlangen, 1848.
Thümmel (A. R.), Neueste Geschichte der Republik Mexiko.
Erlangen, 1848.
Tiempo (El). Mexico, 1834 et seq.
Tilden (Daniel R.), Speech in U. S. House of Rep., July 14, 1846.
Wash., 1846.
Tipografía Mexicana. Mexico, 1878 et seq.
Tlascala, Abusos de los mandones de Indios. MS. Mex., 1629;
Acusacion que la Exma Diputacion Territorial. Mex., 1852;
Despojo á mano armada por el Prefecto de. Puebla, 1843;
Discurso pronunciado el 26 de Dic. de 1851. Puebla, 1852;
Estatuto orgánico del territorio. Mex., 1849; Franciscanos é
Indios. MS.; Representacion que la Diputacion Territorial de.
Mex., 1849.
Todo tiene remedio siendo el aviso oportuno. Mexico, 1822.
Tolerancia de Cultos, Opúsculo que contiene una Corta Defensa
de la Religion de Jesu Cristo. Mexico, 1849.
Tolerancia Religiosa, Disertacion contra la. Mejico, 1831;
Representacion al Sob. Cong. contra el Art. 15 del Proyecto de
constitucion sobre. Mex., 1856.
Toluca, Elecciones de. [Mexico, 1826.]
Tompkins (Patrick W.), Speeches in U. S. House of Rep., Jan. 19
and March 14, 1848. n.pl., n.d.
Tomson, Viage de. In Soc. Mex. Geog., Boletin, 2da Ep., tom. i.
Toral (Francisco), Carta al Adelantado de la Florida. In Cartas de
Indias.
Toral (Francisco), Carta al Rey. In Cartas de Indias.
Tornel (José María), Carta del Gen. á sus amigos. Mex., 1839;
Discurso en la sesion del 12 de Oct. de 1842. Mex., 1842; El
General á sus amigos. Mex., 1839; Fastos militares de
Iniquidad. Mex., 1843; Manifestacion del C. Mex., 1833;
Manifestacion presentada á la Cámara de Sen. Mex., 1841;
Manifiesto del Orígen, Causas, Progresos y Estado de la
Revolucion del Imperio Mexicano. Puebla, 1821; Proyectos de
código criminal y Penal. Mex., 1853; Réplica de Varios
Españoles. Mex., 1841; Respuesta del Gen. Mex., 1840; Varios
Españoles al Sr Tornel, última respuesta. Mex., 1841; Voto
Particular, etc., relativos al Istmo de Tehuantepec. Mex., 1852.
Tornel y Mendivil (J. Julian), La Aparicion de Nuestra Señora de
Guadalupe de Mexico. Orizava, 1849. 2 vols.
Tornel y Mendivil (José María), Breve Reseña Histórica de los
Acontecimientos. Mex., 1852; Decretos, Dec. 29, 1828, Abril 22,
1829, Ag. 26, 1829. Mex., 1828-9; Discurso pronunciado en la
Alameda 27 de Sept. de 1850. Mex., 1850; Discurso que en la
Solemnidad 11 de Set. de 1843. Orizava, 1843; Discurso que al
solemnizarse el feliz natalicio del Emp. Maximilian 6 de Julio de
1864. Orizaba, 1864; Discurso que pronunció el Exmo Sr. Gen.
Mex., 1840; El grito de la patria, Aug. 4, 1821. [Puebla, 1821];
Los Héroes de Dolores vindicados. Mex., 1850;
Proclamaciones, Dec. 27, 1828, Marzo 17, 1829, Abril 28, 1829.
Mex., 1828-9; Protesta del Gen. contra el decreto espedido.
Mex., 1839; Respuesta del Gen. Mex., 1840; Vindicacion del
Coronel. Orizaba, 1842.
Torquemada (Juan de), Monarquía Indiana. Madrid, 1723. 3 vols,
folio.
Torre (Juan de la), Relacion de los Resíduos. In Pacheco and
Cárdenas, Col. Doc., tom. xii.
Torre (Martin de la), Carta sobre el Cometa de 1680. In Morfi, Col.
Doc.
Torrente (Mariano), Historia de la Revolucion Hispano-Americana.
Madrid, 1830. 3 vols.
Torres, Certificacion. In Doc. Hist. Mex., serie ii., tom. iii.
Torres (C. A. de), De Laudibus Mariæ Barbaræ Reginæ 1759.
Mexico, 1760.
Torres (José J. G. de), Desengaño á los Rebeldes sobre su
monstruosa constitucion. [Mex., 1815]; Vindicacion del Clero
Mexicano. Mex., 1812.
Torres (Luis de), Sermon fúnebre que en las honras que hicieron
en 29 de Mayo de 1767. Mexico, 1768.
Torres (Miguel), Vida ejemplar y muerte preciosa. n.pl. [1723] 4to.
Torres del Palacio (Francisco de), Proceso contra. MS., 1595-7.
Torres y Recedo (Marcos), Testamentary disposition, leaving
Government of Mexico to Real Audiencia. MS. [1699.]
Trait D’Union (Le), Mexico, 1861 et seq.
Translationuss hispanischer sprach zü Frantzösisch gemacht so
durch dẽ Vice Rey, etc. 1522.
Trastour (P. E.), Memorial-claim against Tehuantepec Railroad
Company. New-Orleans, 1853.
Trebarra (Napoleon), Los Misterios de Chan Santa Cruz. Mérida,
1834.
Tres Dias de Ministerio, Verdadera Noticia de los. Mexico, 1839.
Tribunal del Consulado. In Soc. Mex. Geog., Boletin, tom. ii.
Tribuno (El), Guadalajara, 1827 et seq.
Tributo á la Verdad. Mexico, 1847.
Tributos. MS.
Tributos, Reglamento de los tributos de esta capital. Mex., 1792;
Reglamento y Ordenanzas. Mex., 1771.
Trigueros (Ignacio), Defensa remitida al Gran Jurado el 13 de Jul.
de 1846. Mexico, 1846.
Triunfo (El), de la justicia en los viles insurgentes. Mexico, 1811.
Trujillo (José Sanchez), Proyecto sobre productos y recursos.
Mexico, 1865.
Tubino (Francisco M.), Un trono en Méjico. Sevilla, 1862.
Tumultos de Mexico. Collection of Documents. MS. folio.
Turchi (Adeodato), Coleccion de los discursos mas interesantes.
Mexico, 1853. 2 vols.
Tyler (Edward B.), Anahuac, or Mexico and the Mexicans. London,
1861.
Ulloa, Mémoires Philosophiques. Paris. 1787.
Ulua (S. Juan de), Proclama. Mexico, 1822.
Unda (P. V.), Sorpresa del campo de Ahuatepec el año de 1828.
Mexico, 1831.
Union de España con la Inglaterra. Mexico, 1809.
United Mexican Mining Association, Report of Directors. London,
1825-1827. 2 vols.
United States Government Documents. [Coast Survey; Commerce
and Navigation; Commercial Relations; Foreign Relations, etc.
Cited by their dates.]
United States Government Documents. [Senate, House,
Miscellaneous, etc. Cited by their congress and session.]
United States and Mexican Boundary Survey. Washington, 1857-
9. 4to. 3 vols.
Universal (El). Mexico, 1849, et seq.
Universidad de Guadalajara, Real Cédula, Deciembre 20, 1815.
Guadalajara, 1816. folio.
Universidad de Mexico, Constituciones. Mex., 1775. folio;
Manifiesto del Ilustre Claustro. Mex., 1810; Obras de eloquencia
premiadas el 28 de Dic. 1790. Mex., 1791; Suplemento á las
Constituciones. Mex., 1839.
Urbano VIII, Breve, acerca de la Ternativa de los Religiosos, etc.
Mexico, 1690.
Uribe (Joseph A. F.), Disertacion Histórica crítica. Mexico, 1801.
Uricoechea (E.), Mapoteca Colombiana. Londres, 1860.
Urizar y Estrada (Juan de), Defensa canónica de la jurisdiccion de
los Prelados Regulares. n.pl., n.d.
Urrea (José), Diario de las Operaciones militares de la division,
etc. Durango, 1838; Protesta contra la violacion de las
capitulaciones de Tampico, etc. Mex., 1839.
Uruena (José Antonio), Defensa de la Secularizacion. Mexico,
1816. MS. 2 vols.
Vacuna (La), Instruccion para ministrar. Mexico, 1814.
Valdés (Joseph Eugenio), Vida Admirable y penitente. Mexico,
1765.
Valdivielso (Ignacio), Apuntes biográficos. Paris, 1861.
Valdovinos (Mucio), Cartilla de las Madres de Familia. Mexico,
1855.
Valencia (Gabriel), Manifiesto sobre su conducta. Mejico, 1841.
Valencia (Obispo de), Representacion á las Córtes. Valencia,
1820.
Valencia et al., Carta al Emperador. In Cartas de Indias.
Valentin (Ph.), Vortrag über den Mexicanischen Calender-Stein.
New York, 1878.
Valladolid, Exhortacion que dirige á los habitantes de la Provincia.
Mex., 1810; Exposicion que el Real Ayuntamiento de la ciudad.
Mérida, 1836.
Valerte (Ignacio L.), La Cuestion de Jalisco. Mexico, 1870.
Valle (J. N. del), El Viajero en Mexico. Mexico, 1859.
Vallecillo Mining Company. New York, 1855.
Vallejo (Josef Ignacio), Vida de la Madre de Dios. Cesena, 1779;
Vida del Señor San José. Mex., 1845.
Vallejo (Mariano G.), Coleccion de Documentos para la Historia de
Mexico. MS. and print. folio. 2 vols.
Valori (Henry de), L’Expédition du Mexique. Paris, 1864.
Valverde (Antonio S.), Idea del valor de la Isla Española. Madrid,
1785.
Vander-Linden (Pedro), Memoria sobre el Cuerpo de Salud Militar.
Mexico, 1845.
Van Dyke (J.), Speech in U. S. House of Rep., Ap. 2, 1848. n.pl.,
n.d.
Vargas (Joseph Mariano de), Carta de pésame por el
fallecimiento. Mexico, 1787. 4to.
Varias Poesías, A Collection.
Variedades, A Collection. 3 vols.
Variedades de la Civilizacion. Mejico, 1852. 3 vols.
Variedades de Jurisprudencia. Mexico, 1850-5. 9 vols.
Variedades del Monitor, A Collection.
Varios Impresos, A Collection. 3 vols.
Varios Papeles, A Collection. 2 vols.
Vattel, Derecho de Gentes ó Principios de la Ley natural Aplicados
á la conducta é Intereses. Burdeos, 1822. 4 vols.
Vazquez, Apuntamiento. In Doc. Inéd., tom. lii.
Vazquez (Rodrigo), Memorial referente á la reparticion y tributo de
las tierras de la Nueva-España. In Pacheco and Cárdenas, Col.
Doc., tom. vi.
Vazquez de Avila (Juan), Carta al Emperador 10 de Febo de 1539.
In Pacheco and Cárdenas, Col. Doc., tom. iii.
Vecinos del Distrito (Los), Solo piden igualdad y justicia en los
impuestos. Mexico, 1852.
Vega (José Sixto González de la), Exhortacion á los Americanos.
Mex., 1811.
Vega (Plácido), Documentos para la Historia de Mexico. MS.,
1862-8. 15 vols.
Vega (Plácido), Da Cuenta al Gobierno de la República Mejicana.
Tepic, 1867; Depósito General de Armas. MS.; República
Mejicana Oficial. n.pl., 1867.
Vega (Rómulo Díaz de la), Manifestacion que hace á los
habitantes de Yucatan. Mérida, 1854.
Vega (S. D. de la), Discurso sobre el objeto de las dramas.
Mexico, 1786.
Veitia Linage, Cuatro Imágenes milagrosas de Nuestra Señora.
MS., 1754.
Vejaciones á los Indios. MS.
Velasco (Alonso Alberto de), Exaltacion de la Divina Misericordia.
Mex., 1699; Mex., 1724; Mex., 1776; Mex., 1807; Historia de la
Milagrosa Renovacion, etc. Mex., 1845; Mex., 1858; Manifiesto
en defensa de la Nota. Mex., 1700.
Velasco (Andrés Miguel Perez de), El Pretendiente de Curatos
instruido. Puebla, 1765.
Velasco (Luiz de), Carta al Rey sobre robo por navíos franceses
en el Puerto de Caballos. MS., 1558.
Velasco y Arellano (Joseph Luis), Amnestria Heróico. Mex., 1711;
Saeta Amorosa. Mex. [1711]; Tierno recuerdo de. Mex., 1726.
Velasco y Texada (Antonio Joseph), Historia de la milagrosíssima
imágen. Mex., 1750; Vozes Jurídicas que en el Tribunal de la
razon de justicia. MS.
Velasquez (Andrés), Carta del á los P. P. Superiores de la Comp.
de Jesus. Puebla, 1748.
Velazquez (Antonio), Instruccion que dió el adelantado á Hernan
Cortés. In Col. Doc. Inéd., tom. xii.
Velazquez de Bazan (Antonio), Memorial del negocio de, cerca de
la merced que pide á su Magestad. In Pacheco and Cárdenas,
Col. Doc., tom. x.
Velazquez (Diego), Carta en la que relaciona la desobediencia de
Hernan Cortés de Octubre 12, 1519. In Pacheco and Cárdenas,
Col. Doc., tom. xii.
Velazco (Francisco de), Hecho del Litigio. Mexico, 1696. MS.
Velez (Pedro), Observaciones sobre el acuerdo de la cámara de
senadores. Mex., 1830; Vindicacion de la primera sala de
suprema corte de Justicia. Mex., 1834.
Venadito (Conde), Document leaving the Government to D. Fran.
Novella, July 5, 1821. Mex., 1821; Fidelísimos habitantes de
esta capital. Mex., 1821; Proclama 3 de Marzo de 1821. Mex.,
1821.
Venégas (Francisco Xavier), Manifiesto de Su Virey á la Nueva
España, Oct. 27, 1810. Mex., 1810; Pronóstico de la Felicidad
Americana. Mex., 1810; Vindicacion de los agravios. Cádiz,
1811.
Venegas (Miguel), Vida y Virtudes del Juan Bautista Zappa.
Barcelona, 1754.
Vera Cruz, Arancel de los Honorarios y Derechos Judiciales. Mex.,
1840; Balanza del Comercio Marítimo, año de 1809. Mex.,
1810, Carta de la Justicia de 10 de Julio, 1519. In Cortés,
Cartas; Carta del Ayuntamiento. In Col. Doc. Inéd., tom. i.;
Documentos referentes á los sucesos ocurridos en Orizava.
Vera Cruz, 1834; El Plan Republicano del Triunvirato. Mex.,
1823; Exposicion de los Empresarios de nuevos Caminos.
Mex., 1834; Esposicion dirigida al Congreso General. Mex.,
1849; Estadística del estado libre y soberano de Jalapa, 1831. 2
vols.; Fábrica de Parroquia en la Antigua. MS. 1783. folio;
Manifiesto del Congreso. Jalapa, 1827; Memoria leida por el
Gobernador del Estado. Vera Cruz, 1871; Memoria presentada
al Hon. Congreso del Estado. Vera Cruz, 1869; Memoria relativa
á la administracion municipal. Vera Cruz, 1865; Oficios del
Consulado. Vera Cruz, 1821; Reglamento de administracion de
propios del Ayuntamiento. Vera Cruz, 1848; Reglamento de
policía y buen gobierno, etc. Vera Cruz, 1821; Reglamento para
la policía interior del Departamento. Jalapa, 1841;
Representacion al Congreso sobre Caminos. Mex., 1835;
Representacion del Vecindario. Vera Cruz, 1821; Sesion del
Hon. Congreso. Vera Cruz, 1833; Sucesos de la actual
revolucion en el Estado. Mex., 1852.
Verdadera Idea sobre el impreso titulado “donativo de medio
millon de Pesos.” Mexico, 1845.
Verdadera Noticia de los Tres Dias de Ministerio. Mexico, 1839.
Verdades de suma importancia á la nacion Mexicana. Mexico,
1826.
Vergara y Estrada (Manuel U.), Eneas Español. Mexico, 1756.
Verona (Paciente de), Paromología de el Diphthongo de
Querétaro en la procesion de el corpus desde el año de 1709.
MS. [1740] 4to.
Vetancvrt (Avgvstin de), Manual de administrar los santos
sacramentos conforme á la reforma de Paula V. y Urbano VIII.
Mexico, 1764.
Vetancvrt (Avgvstin de), Menologio Franciscano de los Varones
mas Señalados. n.pl., n.d. folio; Mexico, 1871.
Vetancvrt (Avgvstin de), Sermon á la aparicion de la milagrosa
imágen de Nuestra Señora de Pilar. Mexico, 1674.
Vetancvrt (Avgvstin de), Teatro Mexicano. Mexico, 1698. folio; also
edition Mexico, 1870-1. 4 vols.
Vetancvrt (Avgvstin de), Tratado de la ciudad de Mexico. n.pl., n.d.
folio.
Veytia (Mariano), Historia Antigua de Mejico. Mexico, 1836. 3 vols.
Veytia (Mariano), Sumaria Relacion de todas las cosas en la
Nueva España y que los tultecas alcanzaron.
Viage á los Estados-Unidos del Norte. Cincinnati, 1834.
Vicarios Generales, Preguntase si ... deban obtener honores de
Ex-Provinciales. n.pl., n.d. 4to.
Victoria (Guadalupe de), Derrotero de las Islas Antillas. Mex.,
1825; Proclama á las provincias de Oriente y Occidente. Mex.,
1823.
Vidal (Joseph), Espada Aguda de Dolor. Mex., 1692; Vida ... del ...
Migvel de Omaña. Mex., 1682.
Vigil (J. M.), and Juan B. de Híjar y Haro, Ensayo Histórico del
Ejército de Occidente. Mexico, 1871-2.
Vigne (G. T.), Travels in Mexico, South America, etc. London,
1863. 2 vols.
Vigneaux (Ernest), Souvenirs d’un Prisonnier de Guerre au
Mexique, 1854-5. Paris, 1863.
Vilaplana (Hermenegildo), Breve Noticia de la portentosa
conversion. Mex., 1760; Histórico y Sagrado Novenario de la
milagrosa imágen, etc. Mex., 1765; Vida portentosa del
Americano, etc. Madrid, 1775.
Villa-Amor (Manuel), Biografía del General Santa Anna. Mexico,
1857.
Villagutierre Soto-Mayor (Juan de), Historia de la Conquista de la
Provincia de el Itza. [Madrid, 1701.] folio.
Villalobos (Gomez de), Relacion del viaje que hizo desde la Nueva
España 1o de Agosto de 1584. In Pacheco and Cárdenas, Col.
Doc., tom. v.
Villalovos (Joachin Antonio de), Relox de sombras. Puebla, 1729.
Villa Manrrique (Marqués de), Carta al Rey Don Felipe II., March
23, 1586. In Cartas de Indias.
Villarroel, Mexico por Dentro y Fuera. Mex., 1831.
Villarroel (Hipólito), Enfermedades políticas. MS. 4 vols.
Villa Señor y Sanchez (José Antonio), Theatro Americano.
Mexico, 1746. folio. 2 vols.
Villavicencio (Juan Joseph de), Vida y Virtudes de el venerable P.
Juan de Ugarte. Mexico, 1752.
Villerias (Joseph de), Llanto de las Estrellas. Mexico, 1725.
Villuendas (Francisco), Estado de la mision de Laredo en 23 de
Noviembre de 1785. MS.
Vindicacion del General Presidente. Mexico, 1837.
Vindicacion del Pueblo Mexicano en la Invasion Francesa de
1862. Mexico. 1867.
Virey Cerralvo, Acusacion contra el, por los Procuradores
generales del Comisario General de la órden de San Francisco
en Nueva España. MS. folio.
Vireyes de Mexico, Instrucciones, Residencias. MSS. and print. 2
vols.
Vives (Juan Luis), Diálogos de Mexico, 1827.
Vivo (Buenaventura), Memorias del ministro de Mejico, en
España, 1853-5. Madrid, 1856; Tratado Consular por el Cónsul
de los Estados-Unidos Mexicanos en la Habana. Mex., 1850.
Vizarron y Eguiarreta, Copia de dos cartas. [Mex., 1742.] MS.
folio; Respuesta y Satisfaccion. Mex., 1772. folio; Sentencia
dada y pronunciada. Mex., 1742. folio.
Vocabulario filosófico-Democrático, Nuevo. Mexico, 1834. 2 vols.
Waldeck (F. de), Voyage Pittoresque, etc. Paris, 1838. folio.
Wappäus (J. E.), Geographie und Statistik von Mexiko und
Central-Amerika. Leipzig, 1863.
War with Mexico, Reports, etc., of Operations of U. S. Naval
Forces, 1846-7 [30th Cong. 2d. Sess. H. Ex. Doc. 1, Pt. ii.].
Washington, 1848.
Ward (H. G.), Gedrängtes Gemälde des Zustandes von Mexiko.
Leipzig, 1828.
Ward (H. G.), Mexico in 1827. London, 1828. 2 vols.
Ward (H. G. Mrs), Six Views of the most important towns and
mining districts of Mexico. London, 1829. folio.
Webster (Daniel), Speech on Mexican War. March 23, 1848.
Washington, 1848.
Webster (Daniel), Speech at Philadelphia, December 2, 1846.
Washington, 1847.
Welda (Othon), Dos Proyectos sobre fundacion de Colonias.
Morelia, 1865.
Welzhofer (Max Moritz), Die Republik Mexico. Leipzig, 1862.
West-Indische Spieghel, door Athanasium Inga. [Amsterdam,
1624.]
Westminster Review. London, 1824 et seq.
Whitehead (Charles), Letter May 11, 1855, to G. W. Lettsom on
debt of Mexican Government to British Bondholders. MS. folio.
Whitehead (Charles), Payment of dividends by Mexican
Government to British Bondholders. May 11, 1855. MS.
Wilhelm (Thomas), History of the Eighth U. S. Infantry, n.pl., 1873.
2 vols.
Willard (Emma), Last Leaves of American History. New York,
1853.
Williams (J. J.), The Isthmus of Tehuantepec. New York, 1852. 2
vols.
Willie (Roberto Crichton), Mexico, Noticia sobre su Hacienda
Pública bajo el Gobierno Español y Despues de la
Independencia. Mexico, 1845.
Willson (Marcius), American history. Cincinnati, 1847.
Wilson (Robert Anderson), A New History of the Conquest of
Mexico. Philadelphia, 1859.
Wilson (Robert Anderson), Mexico and its Religion. New York,
1855.
Wilson (Robert Anderson), Mexico: Its Peasants and its Priests.
New York, etc., 1856.
Winthrop, Speech on the Mexican War, Jan. 8, 1847. Washington,
1847.
Witte (Nicolás), Carta al Emperador, July 15, 1552. In Cartas de
Indias.
Woll (Adrian), Esposicion presentada al Presidente. Mexico, 1852.
Woods (Daniel B.), Sixteen months at the Gold diggings. New
York, 1851.
Wool (John E.), A Sketch of the Life and Public Services of. New
York, 1851.
Worrall (Tomás), Exámen del mérito que puedan tener los
fundamentos, préstamo ... al Supo Gobo. Mexico, 1839.
Worth (W. J.), Charges against by General Scott. n.pl., n.d.
Worthington (C. J.), The Woman in Battle. Hartford, 1876.
Yañez (Mariano), Alegato de buena prueba por parte de la
Compañía Aviadora, etc. Mexico, 1862.
Yañez (Juan) y Socios, Extracto de la causa formado al. Megico,
1839.
Ylarregui (José Salazar), Datos de los trabajos astronómicos y
topográficos 1849-50 por la comision de límites. Mexico, 1850.
Young (Eduardo), Obras Selectas. Mexico, 1833. 4 vols.
Young (Philip), History of Mexico. Cincinnati, 1850.
Yrolo (Nicolás), De las Escripturas. [Mexico, 1605.] 4to.
Yturrigarai (José Manuel), Acto Literario con que la Real y
Pontificia Universidad de Mexico. Mexico, 1803. MS.
Yucatan, Actas de la Junta Electoral de Provincia 12 de Dic. 1830.
Mérida, 1830; Case of H. M. subjects settled on the coast of
Yucatan. Lond., 1789; Compendio histórico razonado del
orígen, etc. Mex., 1825; Dictámen aprobado per la Exma Junta
Departamental. Mérida, 1838; Discurso pronunciado por el
Exmo Señor Gobernador 21 de Agosto de 1840. Mérida, 1849;
Documentos interesantes y decretos del legítimo Cong. Const.
del Estado. n.pl., n.d.; El proyecto de ley del Sr. Senador Vargas
para pacificar el Estado. Mex., 1831; Es el ministerio quien
dirige los planes. Mex., 1831; Estadística. Mex., 1853; Exámen
sobre el actual estado del negocio. Mex., 1832; Exposicion del
Gobierno. Mérida, 1845; Exposicion del Gobierno, derogacion
del decreto 21 Feb., 1844. Mérida, 1844; Exposicion del
Gobierno sobre reconocimiento y amortizacion de los créditos.
Mérida, 1868. 4to; Exposicion que el Real Ayuntamiento.
Mérida, 1836; Hace pocos dias que se publicó. Mex., 1831;
Impugnacion á las observaciones hechas por varios Yucatecos.
Mex., 1831; Impugnacion á las observaciones que en 19 del
Oct. Mex., 1831; Iniciativa de la legislatura del estado. Mérida,
1832; La nueva proposicion del Señor Rejon sobre Yucatan.
Mex., 1831; Las facultades de los supremos poderes federales
en cuanto á la organizacion. Mex., 1832; Manifiesto de la
Convencion del estado. Mérida, 1831; Manifiesto del Congreso
á los habitantes de los estados de la Confederacion Mejicana.
Mérida, 1824; Manifiesto del gefe Superior á los pueblos.
Mérida, 1831; Manifiesto del Gob. Provisional á la nacion
acerca de los negocios. Mex., 1843; Manifestaciones del Exmo
Consejo, ilustres Ayuntamientos. Mérida, 1831; Memoria leida
ante el Augusto Congreso Extraordinario. Mérida, 1846;
Memoria leida ante la Excma Asamblea del Departamento de 7
de Mayo de 1845. Mérida, 1845; Newe zeittung von dem lande
das die Sponier funden haben, 1521. Mex., 1875;
Observaciones sobre el dictámen en la proposicion del Señor
Vargas. Mex., 1831; Observaciones sobre la actual situacion del
departamento. Mex., 1845; Oprimido reclama á la cámara de
Diputados su representacion legal. Mex., 1826; Piezas
justificativas de la conducta política. Mérida, 1846; Proyecto de
Bases para la regeneracion política de la República. Mérida,
1841; Representaciones y Documentos justificativos que la
Legislatura Constitucional. Mérida, 1874; Segunda carta del
Yucateco al Payo de Rosario. Mex., 1829; Tratado que celebró
la administracion provisional con las autoridades. Mex., 1845.
Zacarias (Luis Gonzaga), Representacion que el lic. dirijió al Ilmo
Sr. Obispo de esta Diócesis. Puebla, 1865.
Zacatecas, Acto de un Juicio verbal celebrado en el Juzgado de
distrito. Zac., 1841; Arancel de los honorarios y derechos
judiciales que se han de cobrar. Mex., 1840; Comunicaciones
oficiales entre el Sup. Gob. del Estado y el Superior Ecles. de la
Diócesis. Guad., 1852; Contestacion al folleto titulado “La
Oposicion y el Gobierno.” Aguascalientes, 1853; Continuacion
de las comunicaciones oficiales, Ley de Hacienda. Guad., 1852;
Decreto sobre estabto. de un Banco. Zac., 1830; Dictámen
presentado por la Comision Especial, sobre calificacion de las
Elecciones Nov. 30, 1843. Mex., 1843; Dictámen sobre
Reformas Constitucionales. Mex., 1841; Diario exacto. Mex.,
1835; Impugnacion á la iníciativa del hon. Cong. de ... de 9 de
Jun. de 1834, sobre Reformas Eclesiásticas. Mex., 1834;
Iniciativa de la Honorable Asamblea Departamental sobre
Reformas de las Bases de organizacion política de la
República. Zac., 1845; Memoria presentada por el Ejecutivo á la
Hon. Legislatura. Zac. 1871; Plan para el establecimiento de
una Colonia en el Estado. Baltimore, 1852; Reflecsiones al Hon.
Cong. sobre la esposicion que hace al Sup. Gob. Mex., 1831.
Zaccone (Pedro), El Hijo del Cielo. Mejico, 1858. 2 vols.
Zalatitan, Protesta que el cura y Feligreses de la Parroquia de
Zalatitan hacen contra la Traicion consumada en Veracruz.
Guadalajara, 1860.
Zamacois (Niceto de), El Jarabe, obra de Costumbres Mejicanas.
Mejico, 1861; Historia de Mejico. Barcelona, etc., 1877-80. 11
vols.; Los Ecos de mi Lira. Mex., 1849; Ministerios de Mexico.
Mex., 1850. 2 vols.; Salud del Alma, Devocionario en verso.
Mex., 1851; Un Angel Desterrado del Cielo. Mex., 1855.
Zamboni (J. Fortunato), Ensayo de una memoria sobre la
necesidad de prevenir á los incautos contra los artificios de
algunos fisiólogos. Megico, 1840.
Zambrano (Juan A.), Apuntes sobre Caminos de Fierro y Facilidad
de Hacerlos. Mex., 1867; Comunicaciones Relativas á la
Renuncia que del Cargo de Tesorero General de la Nacion,
hizo. Mex., 1861; Tarifa de Sueldos Militares arreglada para el
uso de los cuerpos de Ejército de la Repúb. Mexicana. Mex.,
1854.
Zamora (Antonio Fernandez), Plan de Hacienda para la República
Mexicana. Mexico, 1847.
Zamora (Victoriano), Informe hecho á la Exma 2a Sala, relativo á
los Ruidosos Autos del Denuncio de San Clemente. Zacatecas,
1839.
Zamora y Azevedo (A. M. and M. D.), Pruebas de Nobleza. MS.
1768.
Zappa (Gio Battista), Copia di Vintisei Lettere Scritte dal Padre al
Sig. Ambrosio Zappa. Milano [1687].

Artificial Intelligence and Machine Learning in the Travel Industry: Simplifying Complex Decision Making Ben Vinod full chapter instant download

Uploaded by

Copyright:

Available Formats

You might also like

Artificial Intelligence and Machine Learning in the Travel Industry: Simplifying Complex Decision Making Ben Vinod full chapter instant download

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Artificial Intelligence and Machine Learning in the Travel Industry: Simplifying Complex Decision Making Ben Vinod full chapter instant download

Uploaded by

Copyright:

Available Formats

Artificial Intelligence and Machine

Learning in the Travel Industry:

Analytics Enabled Decision Making Vinod Sharma

Risk Modeling: Practical Applications of Artificial

Artificial Intelligence and Machine Learning for EDGE

Sustainable Developments by Artificial Intelligence and

Machine Learning for Decision Makers: Cognitive

Machine Learning for Decision Makers: Cognitive

Predicting Heart Failure : Invasive, Non-Invasive,

Automation and Autonomy: Labour, Capital and Machines

Artificial Intelligence and Machine

Previously published in Revenue and Pricing Management

Special issue on artificial intelligence/machine learning in travel ..................................................................................... 1

Machine learning approach to market behavior estimation with applications

Special issue on artificial intelligence/machine learning in travel

Published online: 13 March 2021

Reprinted from the journal 1

2 Reprinted from the journal

Price elasticity estimation for deep learning‑based choice models:

Introduction performing tasks such as demand modeling and assortment

Reprinted from the journal 3

4 Reprinted from the journal

Elasticity Compared to other industries (e.g., retail), the study

Reprinted from the journal 5

direction using some step size. For an in-depth treatment

State‑of‑the‑art deep network‑based choice models

Alternatives and individuals making choices can be

6 Reprinted from the journal

Fig. 3 The methodology applied

Reprinted from the journal 7

Fig. 4 Performance of different

8 Reprinted from the journal

Reprinted from the journal 9

sensitivity, while in business trips we see a significant differ-

10 Reprinted from the journal

Reprinted from the journal 11

can be inelastic or not. Some cases where we observed

12 Reprinted from the journal

Fig. 10 Median elasticity (neg)

Reprinted from the journal 13

Table 2 Trip purpose experiments and results on hold-out test set

Full feature set H2O GBM 48 86.5

validation set to estimate the depth of the individual trees via

14 Reprinted from the journal

You might also like