China missiles, China ballistic missiles, China missile technology ...

hulod - Stephane Hessinstitute For Transport Studies , University Of Leeds

Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres

44 pages

English

Le téléchargement nécessite un accès à la bibliothèque YouScribe
Tout savoir sur nos offres

A propos
Informations
Extrait

Description

cours - matière potentielle : missile defenses
mémoire
exposé
mémoire - matière potentielle : sfw
expression écrite

China's Evolving Conventional Strategic Strike Capability The anti-ship ballistic missile challenge to U.S. maritime operations in the Western Pacific and beyond Mark Stokes September 14 th 2009

asbm
strategic strike capability
sar systems iir
synthetic aperture radar sast
ballistic missile challenge to u.s.
entry vehicle from the extreme heat
space

Sujets

Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Hybrid choice models
Stephane Hess
Institute for Transport Studies, University of Leeds
Danish Choice Modelling Day, Odense, December 2011
1 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Presentation outline
1 Introduction
2 Hybrid models
3 Case study I: attitudes
4 Case study II: APS
5 Case study III: scale
6 Conclusions
2 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Introduction I
Choice models explain decisions as a function of sensitivities to
different attributes describing the alternatives
Aim is to better understand individual sensitivities
Analysts are justiﬁably keen to make use of any additional
information/data available to them
Partly a reﬂection of the notion that sensitivities that drive choice
behaviour also play a role in other decisions, answers, etc
3 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Introduction II
Some examples:
Respondents’ answers to attitudinal questions
e.g. “On a scale from 1 to 5, to what extent do you agree with ..." answers to attribute processing questions
non-attendance, ranking, etc
Respondents’ answers to questions relating to survey complexity,
as well as information on survey completion time
Observations relating to other choices made by the same
respondent
e.g. respondents indicates that he/she travels by rail because it is
more environmentally friendly
4 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Introduction III
Every single one of these has been used in past work, through
interactions in the utility function
Often leading to important gains in model ﬁt, and retrieving
reasonable effects
Unfortunately, this is all rather unwise
5 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Endogeneity
Let’s take a look back at the theory
Utility that respondent n obtains from alternative i in choice
situation t is given by U = V +"i;n;t i;n;t i;n;t
Made up of a deterministic and random component
A key assumption is that of independence between the
deterministic utility component and the random utility component
The deterministic approaches above use the additional
information directly in the utility function
Answers to attitudinal questions, attribute processing questions,
etc, are likely linked to sensitivities, so could help explain
heterogeneity
But they are also likely correlated with other effects that are not
captured in our utility function
This correlation between V and" can lead to endogeneityi;n;t i;n;t
bias
6 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Measurement error, and forecasting
Answers to attitudinal questions are not direct measures of
attitudes but functions of underlying latent attitudes
Same thing applies for many of the other components commonly
used
This can lead to measurement error
Also not applicable for forecasting (general SP issues aside)!
7 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Hybrid models
Recognise that attitudes, processing strategies, survey
engagement etc are not observed (i.e. are latent), and indicators
should thus not be treated as explanatory variables, but as
dependent variables
Formulate a structure that jointly models the observed/stated
choice and the values of these indicators (e.g. response to
attitudinal questions)
The link between the two model components is made by latent
variables relating to the underlying attitudes, sensitivities,
processing strategies, etc
Signiﬁcant developments in e.g. Ben-Akiva et al. (2002); Bolduc
et al. (2005) in the context of attitudes
Ben-Akiva, M., Walker, J., Bernardino, A., Gopinath, D., Morikawa, T., Polydoropoulou, A., 2002. Integration of
choice and latent variable models. In: Mahmassani, H. (Ed.), In Perpetual motion: Travel behaviour research
opportunities and application challenges. Pergamon, Ch. 13, pp. 431–470.
Bolduc, D., Ben-Akiva, M., Walker, J., Michaud, A., 2005. Hybrid choice models with logit kernel: Applicability to
large scale models. In: Lee-Gosselin, M., Doherty, S. (Eds.), Integrated Land-Use and Transportation Models:
Behavioural Foundations. Elsevier, Oxford, pp. 275–302.
8 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Choice model
9 / 44Introduction Hybrid models Case study I: attitudes Case study II: APS Case study III: scale Conclusions
Using indicators as explanators
10 / 44