Computer Science
5 years ago

New Prior Method Simplifies Bayesian Model Selection for Linear Models

arXiv (Cornell University)
Power-Expected-Posterior Priors as Mixtures of g-Priors
Dimitris Fouskakis, Ioannis Ntzoufras

Paper Summary

The article discusses a new method called the power-expected posterior prior, which helps in choosing the best model in statistics. This method is a mix of different types of prior beliefs, making it easier to calculate results. The researchers found that this method can be used to compare different models and make better decisions based on the data.

Related papers

Theme background image
Computer Science
8 years ago

New prior distributions lead to more stable and accurate predictions.

Model-Based Clustering
Regularization Methods
Missing Data
Theme background image
Mathematics
8 years ago

New method revolutionizes model selection in statistical analysis for better decisions.

Missing Data
Skew Distributions
Model-Based Clustering
Theme background image
Mathematics
8 years ago

New method ensures accurate data analysis for complex statistical models.

Skew Distributions
Missing Data
Model-Based Clustering