Hierarchical Dirichlet process
 | This article needs attention from an expert in Statistics. Please add a reason or a talk parameter to this template to explain the issue with the article. (February 2012) |
 |
| This article needs additional citations for verification. (February 2012) |
In statistics, the hierarchical Dirichlet process is a nonparametric Bayesian approach to modeling grouped data. It uses a Dirichlet process, whose base distribution is itself drawn from a Dirichlet process. This method allows clusters to share statistical strength. If a single Dirichlet process is used instead, no sharing can occur between groups of data as the atoms drawn for each group will be different with probability one. However, if the base distribution is itself distributed as a Dirichlet process, it will have a discrete support, allowing the groups to share atoms.[1]
References
- ^ Teh, Y. W.; Jordan, M. I.; Beal, M. J.; Blei, D. M. (2006). "Hierarchical Dirichlet Processes" (PDF). Journal of the American Statistical Association. 101: pp. 1566–1581.