Printable Handouts

PDF

Navigable Slide Index

1. Introduction
2. Talk outline
3. Bayesian theory and informative priors
4. Characterizing the Bayesian approach
5. Visualizing the prior
6. Subjectivity and prior beliefs
7. Where does the prior come from?
8. Statistical models for making pricing decisions
9. Goal
10. Weekly movement and price of TropPrem64
11. Movement vs. price of TropPrem64
12. Movement vs. price of TropPrem64 - log plot
13. Sales response models
14. Applications of this model
15. Hierarchical Bayesian models
16. The problem with the usual regression approach
17. Our solution: shrinkage
18. Rewriting our model in vector notation
19. A Bayesian version of our model
20. Hierarchical Bayesian setup
21. Our Bayesian setup
22. Implication: shrinkage estimators
23. What are we shrinking towards?
24. Illustrating how shrinkage works
25. Out-of-sample predictions
26. Store-level strategies
27. Movement vs. price in two Chicago stores
28. Profitability of TropPrem64 in two stores
29. Store-level national brand / store brand price gaps
30. Findings
31. Theory based priors
32. Informative priors
33. A typical sales response model
34. Additive utility structure
35. A prior based upon a simple additive utility model
36. Our approach to specifying the prior
37. A simulated example
38. Implementation
39. Adaptive shrinkage
40. Demonstrating adaptivity with a simulation
41. Application to retailer scanner data
42. Comparison of goodness-of-fit
43. Goodness-of-fit for various estimators
44. Marginal prior distributions
45. Elasticity estimates - unrestricted least squares
46. Elasticity estimates - Bayes
47. Conclusions (1)
48. Eliciting priors using experts
49. Constraints as business rules
50. Constraints represent prior information
51. Example
52. Practical problem
53. Log-linear demand example
54. Computing the implied prior (1)
55. Computing the implied prior (2)
56. Visualizing the implied prior
57. Solution
58. Another example
59. Implied prior
60. Distribution of optimal prices for various priors
61. Empirical application
62. Rejection sampling
63. Implied prior of own-price on TropPrem64
64. Impact on the posterior of own-price coefficients
65. Posterior distribution of optimal price
66. Conclusions (2)
67. Final recommendations
68. Suggestions

Topics Covered

• 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

Links

Series:

• Bayesian Analysis in Marketing

Categories:

• Marketing & Sales

Talk Citation

Publication History

• Published on January 28, 2010