Combining experts' opinions using a normal-Wishart model

Article Abstract:

A study exploring how the combination of experts' opinions by using a normal-Wishart model affects a decision-maker's forecast on the outcome of specific variables was conducted. There are a number of inherent problems with the use of normal-Wishart assumptions on the model generated by the study. These include the inconsistency of implied likelihood functions with prior knowledge and the model's deficiency when different amounts of information are presented by more than one expert. However, the model compared favorably to an updating procedure, when using read data sets.

Author: Wiper, Michael P., French, Simon
Decision-making, Decision making, Specialists, Specialization

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


Hierarchical Bayes forecasts of multinomial Dirichlet data applied to coupon redemptions

Article Abstract:

A method to forecast the monthly and total redemptions of coupons was developed. The procedure used the multinomial Direchlet distribution and hierarchical Bayes (HB) methods to determine the rate of redemptions of off-cents promotional coupons over an 18-month period. Results showed that the HB forecasts were similar to the pooled forecasts. However, the HB model gives a realistic estimate of standard errors. In addition, the HB forecasts were more accurate and adapt more readily than the pooled forecasts.

Author: Lenk, Peter J.
Forecasts and trends, Coupons (Retail), Discount coupons, Series, Dirichlet, Dirichlet series

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


BVAR as a category management tool: an illustration and comparison with alternative techniques

Article Abstract:

The use of Bayesian vector autoregression for simplifying the decision task of a product category manager is discussed. The modeling approach is especially suited to marketing because it accurately predicts a category's state variables, encapsulates managerial actions and provides impulse response analyses. This was confirmed in a comparison of BVAR with Box-Jenkins transfer function analysis and autoregressive moving-average modeling.

Author: Mathur, Sharat K., Curry, David J., Whiteman, Charles H., Divakar, Suresh
Models, Product management, Autoregression (Statistics), Prediction theory

User Contributions:

Comment about this article or add new information about this topic:

CAPTCHA


Subjects list: Research, Bayesian statistical decision theory, Bayesian analysis
This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.