On exact Bayesian credible sets for discrete parameters

Area of Science:

• Statistics
• Bayesian Inference
• Decision Theory

Background:

• Traditional Bayesian credible sets have limitations in achieving specific probability levels.
• Existing methods may not offer flexibility in setting desired credible levels.

Purpose of the Study:

• To introduce a generalized Bayesian credible set.
• To enable the achievement of any preassigned credible level.
• To address limitations in current credible set methodologies.

Main Methods:

• Exploiting the connection between highest posterior density sets and the Neyman-Pearson lemma.
• Developing a generalized framework for Bayesian credible set construction.

Main Results:

• A novel generalized Bayesian credible set is introduced.
• The method allows for the preassignment of any desired credible level.
• Demonstrated the effectiveness of the approach in enhancing Bayesian analysis.

Conclusions:

• The generalized Bayesian credible set offers enhanced flexibility and precision.
• This advancement overcomes limitations of existing credible sets.
• The findings have implications for robust statistical inference and decision-making.