Exploring dependence between categorical variables: Benefits and limitations of using variable selection within Bayesian clustering in relation to log-linear modelling with interaction terms
详细信息 查看全文
- 作者：Michail Papathomasa ; M.Papathomas@st-andrews.ac.uk" class="auth_mail" title="E-mail the corresponding author ; Sylvia Richardsonb
- 关键词：Bayesian model selection ; Sparse contingency tables ; Graphical models
- 刊名：Journal of Statistical Planning and Inference
- 出版年：2016
- 出版时间：June 2016
- 年：2016
- 卷：173
- 期：Complete
- 页码：47-63
- 全文大小：1128 K
文摘
- •
We relate two approaches; Bayesian partitioning and log-linear modelling.
- •
We derive theoretical results on this relation, plus results based on simulations.
- •
Illustrations show that partitioning can assist log-linear model search.
- •
Detecting marginally independent covariates assists the search for interactions.
- •
The main advantage concerns sparse contingency tables.