Supervised MELD for Multi-domain Mixed Membership Analyses
| dc.contributor.advisor | Dunson, David B | |
| dc.contributor.author | Shao, Wei | |
| dc.date.accessioned | 2017-08-16T18:26:26Z | |
| dc.date.available | 2017-08-16T18:26:26Z | |
| dc.date.issued | 2017 | |
| dc.department | Statistical Science | |
| dc.description.abstract | When variables used in a mixed membership analysis can be classied into conceptu- ally distinct domains, interpretation of results is facilitated by using domain-specic models with a small number of imposed pure-type proles and yielding a set of Grade- of-Membership scores for each domain. We present Supervised Moment Estimation of Latent Dirichlet Models (supervised MELD) algorithms for mixed membership models to be used and tested in this context. We challenge the methodology with data sets collected over time to study malaria risk on the Brazilian Amazon frontier. By further ignoring spatial specicity and utilizing a binary outcome variable (at least one person in a household infected with malaria during the past year), we nd supervised MELD capable of extracting surprisingly nuanced risk patterns. | |
| dc.identifier.uri | ||
| dc.subject | Statistics | |
| dc.subject | Feature selection | |
| dc.subject | Machine learning | |
| dc.subject | Supervised learning | |
| dc.title | Supervised MELD for Multi-domain Mixed Membership Analyses | |
| dc.type | Master's thesis |
Files
Original bundle
- Name:
- Shao_duke_0066N_14098.pdf
- Size:
- 330.33 KB
- Format:
- Adobe Portable Document Format