Henry Stewart Talks

The Marketing & Management Collection

Online Seminars By Leading World Experts
Search

Talk: Introduction to MCMC methods and the Gibbs sampler (28 min)

 Loading...

This talk is not yet available, however should be within a few days.

We apologize for the inconvenience.

- Henry Stewart Talks

We hope you have enjoyed this limited-length demo talk

  • Review options for obtaining full talk access by clicking 'Purchase Options' on the right.
  • Would you like to request to be kept updated on The Marketing & Management Collection?
Keep Me Updated       Replay Talk
Close This is a limited length demo talk; you may login or review purchase options.

SHARE THIS TALK

Email via Local Program (e.g. Outlook) Email via the Web Share with Facebook Tweet on Twitter Share with Gmail Share with VKontakte Share with LinkedIn More

DETAILED SLIDE INDEX

  1. 1. Introduction
  2. 2. Simulation methods
  3. 3. MCMC methods
  4. 4. The Markov chain
  5. 5. Ergodicity (1)
  6. 6. Practical considerations
  7. 7. Asymptotics
  8. 8. Simulating from Bivariate normal
  9. 9. Gibbs sampler (1)
  10. 10. Gibbs sampler (2)
  11. 11. Hammersley-Clifford theorem
  12. 12. rbiNormGibbs (1)
  13. 13. Intuition for dependence
  14. 14. rbiNormGibbs (2)
  15. 15. Ergodicity (2)
  16. 16. Relative numerical efficiency (1)
  17. 17. Relative numerical efficiency (2)
  18. 18. General Gibbs sampler
  19. 19. Different prior for Bayes regression
  20. 20. Different posterior
  21. 21. Different simulation strategy
  22. 22. runiregGibbs (1)
  23. 23. runiregGibbs (2)
  24. 24. runiregGibbs (3)
  25. 25. runiregGibbs (4)
  26. 26. R session (1)
  27. 27. R session (2)
  28. 28. R session (3)
  29. 29. Draws of beta
  30. 30. Draws of sigma squared
  31. 31. Summary
  32. 32. END

RELATED TALKS

Play 'Issuance methods of government securities'
Issuance methods of government securities More info
Dr. David Goldreich

TOPICS COVERED IN THIS TALK

  • The primary market for treasury securities
  • Alternative treasury auction mechanisms and the winner's curse
  • Auction success and comparison of mechanisms
  • Key questions in auction design
Play 'Introduction to the series'
Introduction to the series More info
Mr. Andrew Ingram

TOPICS COVERED IN THIS TALK

  • Introduction to "Getting the Most out of Radio Advertising"
Play 'An introduction to managing people in organizations'
An introduction to managing people in organizations More info
Dr. Tyrone Pitsis

TOPICS COVERED IN THIS TALK

  • Overview
  • Shifting emphasis
  • Orientation
  • Reality in management?
  • The nature of management
  • Managing people
  • The brain metaphor
  • Uncertainty
  • Perception
  • Perceptual errors
  • Affective forecasting
  • Key pointers
  • Managing perceptual errors
  • Positive psychology
  • Learned helplessness
  • Helplessness vs. helpfulness
  • Application (appreciative inquiry)
  • 24 core strengths
  • Strengths psychology
Play 'Building permission-based relationships: new methods for gaining opt-in'
Building permission-based relationships: new methods for gaining opt-in More info
Mr. Michael Hanley
Mr. Gerry Christensen
Ms. Diane Strahan
Mr. Chris Torbit
Mr. James Darcey
Dr. G.D. Ramkumar

TOPICS COVERED IN THIS TALK

  • Zoove StarStar (**) method of opt-in
  • Evolution of direct response
  • State of mobile marketing
  • Direct response business model
  • Benefits
  • Common short codes
  • What is the best way to mobilize your marketing campaigns?
  • How does mobile marketing work for brands?
  • Uses for common short codes
  • IVR
  • How can IVR be used in mobile marketing programs?
  • Voice is simple
  • Double opt-in and IVR
  • Image recognition based opt-in for mobile marketing
  • Advantages of image recognition opt-in
  • Image recognition technology
Play 'Discourse analysis as a method in strategy-as-practice'
Discourse analysis as a method in strategy-as-practice More info
Prof. Eero Vaara

TOPICS COVERED IN THIS TALK

  • Discursive perspectives on strategy
  • Strategy discourse
  • Narratives
  • Conversations
  • Strategic plans
  • Media texts
  • Discursive analysis of strategy
  • Strategy as multifaceted discourse
  • Interdiscursivity and recontextualization
  • Discourse and power in context
  • Discursive struggles
  • Discourse and materiality
  • Methodological questions

EMBED THIS TALK

Embed this talk in a Word document   Word       Embed this talk into your website   HTML      

TALK'S CITATION

Rossi, P. (2010), "Introduction to MCMC methods and the Gibbs sampler", in Allenby, G.M. and Rossi, P.E. (eds), Bayesian Analysis in Marketing: A breakthrough in customer analytics, The Marketing & Management Collection, Henry Stewart Talks Ltd, London (online at http://hstalks.com/?t=MM0992423)

RELATED SERIES

Sport Marketing
Analysis of all the essential aspects of sport marketing, illustrated with numerous real world case studies
Financial Services Marketing
The role and power of marketing in the financial environment
Marketing Hotels
Addressing the unique challenges of marketing hotels
Mobile Marketing
How to leverage the mobile channel for marketing
Retail Marketing
The marketing issues facing a fast-changing sector
Welcome, Guest
Cart (0)   Login  
Free Institution-Wide Trial Recommend to Librarian Keep Me Updated Purchase Options

Purchase Individual Talk

24-hour online access for only $29.99
More info

TALK PURCHASE INFORMATION

Immediate online access to an Individual Talk is provided with a Single User License for a period of 24 hours.

The 24-hour period will commence after purchase completion, and from the first time you click to play the purchased talk. During the 24-hour period you have unlimited, full access to the purchased talk.

Price: $29.99

Add Talk to Cart

Purchase Series

3-year online access for only $569.99
More info

SERIES PURCHASE INFORMATION

Immediate online access to this Full Series of 23 talks is provided with a Single User License for a period of 3 years from purchase completion.

Price: $569.99

Add Series to Cart

Obtain Institutional License

Obtain a custom license for your institution

ABOUT THIS TALK

Speaker(s)

Prof. Peter Rossi Show Biography

SPEAKER BIOGRAPHY

Prof. Peter Rossi – James Collins Professor of Marketing at the Anderson School of Management, UCLA, USA

Peter Rossi is the James Collins Professor of Marketing at the Anderson School of Management at UCLA. He received his PhD from University of Chicago and BA from Oberlin College. He has published widely in marketing, economics, statistics and econometrics including Quantitative Marketing and Economics, Marketing Science, Journal of Marketing Research, American Economic Review, Journal of the American Statistical Association, Econometrica, Journal of Political Economy, Journal of Econometrics, Biometrika, Journal of Business and Economic Statistics, Rand Journal of Economics, and Journal of Economic Theory. These articles have more than 5600 Google Scholar cites. He is a co-author of Bayesian Statistics and Marketing, John Wiley Series in Probability and Statistics (2005). Professor Rossi founded the Kilts Center for Marketing, Booth School of Business, University of Chicago while on faculty there. His areas of research interest include pricing and promotion, target marketing, direct marketing, micro-marketing, limited dependent variable models and Bayesian statistical methods.

Publication Date

January, 2010

Topics Covered

MCMC methods... more

TOPICS COVERED IN THIS TALK

  • MCMC methods
  • Ergodicity
  • Gibbs sampler
  • Relative numerical efficiency
  • Bivariate normal example
  • A different prior for Bayes regression

Series

Bayesian Analysis in Marketing

OTHER TALKS IN THIS SERIES

FUNDAMENTALS
Play '1. Bayesian essentials and bayesian regression'
1. Bayesian essentials and bayesian regression More info
Prof. Peter Rossi

TOPICS COVERED IN THIS TALK

  • The goal of inference
  • Bayes theorem
  • Decision theory
  • Sampling properties of Bayes estimators
  • Costs and benefits of Bayes
  • Conjugate families
  • Beta-binomial model
  • Bayesian regression
  • IID simulator
  • Shrinkage and conjugate priors
Now Playing
2. Introduction to MCMC methods and the Gibbs sampler
Prof. Peter Rossi
Play '3. Hierarchical models, conditional independence and data augmentation'
3. Hierarchical models, conditional independence and data augmentation More info
Prof. Greg M. Allenby

TOPICS COVERED IN THIS TALK

  • Fly fishing: a hierarchical model
  • Motivation
  • Traditional estimation
  • Gibbs sampler for the probit model
  • Probit conditional distributions
  • Inverse cdf
  • Basics of hierarchical models
  • Conditional independence
  • Mixtures of normals
  • Model hierarchy
  • Gibbs sampler for mixtures of normals
  • Multivariate mix of norms Ex
Play '4. Metropolis algorithms, logit and quantile regression estimation'
4. Metropolis algorithms, logit and quantile regression estimation More info
Prof. Greg M. Allenby

TOPICS COVERED IN THIS TALK

  • Review of Markov chains
  • Stationary distribution
  • Time reversible chains
  • Metropolis methods
  • Discrete case: Metropolis-Hastings algorithm
  • Independence chains
  • Random walk chains
  • Relative numerical efficiency
  • The Gibbs sampler
  • Logit model
  • Scaling rw metropolis
  • Quantile regression
  • Asymmetric Laplace distribution
Play '5. Unit-level models and discrete demand'
5. Unit-level models and discrete demand More info
Prof. Greg M. Allenby

TOPICS COVERED IN THIS TALK

  • Economic models
  • Logit and probit models
  • Linear utility, contours and budget
  • Choice probabilities
  • Multinomial probit models
  • Differenced system
  • Conditional normal distribution
  • Identification issues
  • Superior and normal goods
  • Displacement parameters
  • Kuhn-Tucker conditions for satiation
  • Estimating the multivariate probit model
  • Component loadings
Play '6. Heterogeneity'
6. Heterogeneity More info
Prof. Greg M. Allenby

TOPICS COVERED IN THIS TALK

  • Panel structure data
  • Heterogeneity and priors
  • Hierarchical linear model
  • Priors
  • Adaptive shrinkage
  • Heterogeneous logit model
  • Random effects with regressors
  • Heterogeneous logit
  • Running rhierBinLogit
  • Data(bank)
  • Sample observations
  • Distribution of heterogeneity for selected part-worths
  • Extensions
Play '7. Model choice and decision theory'
7. Model choice and decision theory More info
Prof. Peter Rossi

TOPICS COVERED IN THIS TALK

  • Model selection
  • Bayes factors
  • Asymptotic methods
  • Newton raftery
  • Marketing decisions and BDT
  • Lost and profit functions
  • Why full Bayes?
  • Value of disaggregate information
  • Value of household purchase information
  • Couponing example
Play '8. Bayesian instrumental variables and simultaneity'
8. Bayesian instrumental variables and simultaneity More info
Prof. Peter Rossi

TOPICS COVERED IN THIS TALK

  • Simultaneity and the marketing mix
  • Conditional likelihood
  • Instrumental variables
  • The linear case
  • Identification
  • Priors
  • Gibbs sampler
  • Non-random marketing mix Variables
  • The sales force allocation problem
  • NBD regression
  • The marginal distribution of X
  • The full model
APPLICATIONS
Play '9. The value of HB in conjoint/choice analysis'
9. The value of HB in conjoint/choice analysis More info
Mr. Bryan K. Orme

TOPICS COVERED IN THIS TALK

  • Introduction to conjoint/choice analysis
  • How HB has improved ratings-based conjoint methods
  • Why HB has particularly benefited choice-based models
  • Market simulation issues including IIA
  • Practical considerations such as utility constraints and using proper priors
Play '10. The SoV Probit'
10. The SoV Probit More info
Dr. Jeff Brazell

TOPICS COVERED IN THIS TALK

  • Source of volume and cannibalization
  • Product and line optimization questions
  • Quick definitions
  • Red bus / blue bus example
  • What is IID? and IIA?
  • We know IIA/IID gets source-of-volume wrong
  • When should we care about substitution patterns?
  • Source of volume calculations
  • Cannibalization = Preference
  • How do you "handle" IIA in your choice models?
  • Simulating preference shares: state-of-the-art
  • What does proportional substitution look like?
  • Industry survey: how do current methods do?
  • The SoV Probit
  • Why should I care about correlated errors?
  • Why aren't we all using Probit models?
  • A structured covariance Probit
  • Distance metrics
  • Choice experiment
  • Model fit statistics
  • Parameter estimates
  • WTP comparison
  • Representative correlation matrix
  • Does it work?
Play '11. Trade-up and trade-down: models for capturing change in consideration set'
11. Trade-up and trade-down: models for capturing change in consideration set More info
Mr. Mark Garratt

TOPICS COVERED IN THIS TALK

  • Pricing issues that arise in categories where there are differences in quality
  • Introduction to a methodology for understanding budget constraints, price-value of product attributes and price power of superior goods
  • Case studies in health and beauty, and consumer electronics categories
  • Comparison to results obtained from standard choice and conjoint methods
Play '12. Bayesian modeling of social network data'
12. Bayesian modeling of social network data More info
Prof. Asim Ansari

TOPICS COVERED IN THIS TALK

  • Social networks
  • Some modeling questions and issues in social network research
  • Nature of network data
  • Network structure and heterogeneity
  • Statistical models of network structure
  • Sender and receiver effect
  • Selectivity and reciprocity
  • Bayesian estimation
  • Schema of a social network
  • Swiss online social network of music
Play '13. Joint choice decisions'
13. Joint choice decisions More info
Prof. Neeraj Arora

TOPICS COVERED IN THIS TALK

  • A novel methodology to estimate joint preference
  • No need to collect joint data, individual data is enough
  • Merges experimental design, statistical modeling and utility aggregation theories to measure joint preference
  • Identifies serious limitations of individual preference as a surrogate for joint preference
  • An empirical application involving cell phone purchase by parent teen dyads
  • Applies to B2B domains
Play '14. Estimating an item's category role'
14. Estimating an item's category role More info
Dr. Peter Boatwright

TOPICS COVERED IN THIS TALK

  • Product categories introduced
  • How individual items influence category sales
  • Overview of data structure as lead-in to model
  • Model development
  • Model estimates and results
Play '15. Bayesian stochastic dynamic models for internet auctions'
15. Bayesian stochastic dynamic models for internet auctions More info
Prof. Eric T. Bradlow

TOPICS COVERED IN THIS TALK

  • Our integrated stochastic model approach
  • Data description
  • Empirical findings
  • Conceptual model development
  • Latent bidders
  • Model development
  • Model results
  • Model fit
Play '16. Hierarchical effects of advertising'
16. Hierarchical effects of advertising More info
Prof. Sandeep R. Chandukala

TOPICS COVERED IN THIS TALK

  • Multistage processes
  • Understanding intermediate effects of advertising
  • A simple Bayesian approach for capturing intermediate effects with heterogeneous response segments
  • Application based on cross-sectional brand tracking data to demonstrate the proposed modeling strategy and model comparisons
Play '17. Investigating the strategic influence of customer satisfaction on firm financial performance'
17. Investigating the strategic influence of customer satisfaction on firm financial performance More info
Dr. Jeffrey Dotson

TOPICS COVERED IN THIS TALK

  • Formal definition of strategic and tactical marketing decisions
  • Linking customer and employee satisfaction to firm performance
  • Models of simultaneous supply and demand
  • Approaches to optimal allocation of marketing resources
Play '18. A Bayesian approach to attribute based consideration sets'
18. A Bayesian approach to attribute based consideration sets More info
Prof. Timothy J. Gilbride

TOPICS COVERED IN THIS TALK

  • The consumer decision problem and two stage decision rules
  • Discrete choice models and consideration sets
  • Using data augmentation and MCMC methods to model consideration sets
  • Alternative screening rules and how to operationalize them
  • An empirical application
  • Summary and extensions
Play '19. Variety: models of multiple-discreteness'
19. Variety: models of multiple-discreteness More info
Prof. Jaehwan Kim

TOPICS COVERED IN THIS TALK

  • Issue of multi-item buying behavior in scanner-panel data
  • Preference and satiation in consumer behavior
  • Economic theory for multiple-discreteness and challenges in model building
  • How to build models at product offering level
  • How to build models at product characteristic level
  • Application of models to marketing decision
  • Discussion on future direction for model extension
Play '20. Models for upper levels of a hierarchy'
20. Models for upper levels of a hierarchy More info
Dr. Qing Liu

TOPICS COVERED IN THIS TALK

  • Hierarchical Bayes models
  • The level-effect problem
  • Contextual influence
  • Range-frequency theory
  • Combining with distance-density theory
  • Model formalization
  • Web-based surveys
  • In-sample fit model comparison
  • Future research
Play '21. Making better pricing decisions with informative priors'
21. Making better pricing decisions with informative priors More info
Dr. Alan L. Montgomery

TOPICS COVERED IN THIS TALK

  • Bayesian theory and informative priors
  • Using statistical models for making pricing decisions
  • Sales response models
  • Exchangeable priors: hierarchical Bayesian models
  • Implication: shrinkage estimators
  • Store-level strategies
  • Theory based priors
  • Informative priors
  • Additive utility structure
  • A simulated example
  • Elasticity estimates
  • Eliciting priors using experts
  • Constraints as business rules
  • Empirical application
  • Rejecting sampling
  • Final recommendations
Play '22. Marketing mix modeling'
22. Marketing mix modeling More info
Prof. Thomas Otter

TOPICS COVERED IN THIS TALK

  • The role of managers
  • What about standard tools?
  • Simple example
  • More realistic example
  • Simulation study
  • gains from reallocation
  • Modeling challenges
  • Modeling strategy
  • Estimation strategy
  • Testing for the supply-side model
  • Non-informative versus flat priors
  • Predictive test
  • Application
  • Analysis
Play '23. Reporting bias in survey data'
23. Reporting bias in survey data More info
Dr. Sha Yang

TOPICS COVERED IN THIS TALK

  • Survey method
  • A general modeling framework
  • Longitudinal panel survey data
  • The challenge
  • The proposed model
  • Modeling true behavior
  • The HB estimation framework
  • Include state dependence
  • The other model parameters
  • Simulation based on cross sectional data
  • An empirical application
  • Possible extensions

We use cookies on this site to enhance your user experience

By clicking any link on this page you are giving your consent for us to set cookies.