Alexandrou-Majaj, Nicolas (2026) Poisson vs Negative Binomial GLMs for Modeling Hospital Length of Stay. Bachelor's Thesis, Applied Mathematics.
|
Text
BachelorsProjectDRAFTFINAL.pdf Download (2MB) | Preview |
Abstract
This thesis investigates the modeling of hospital length of stay using count regression models. The main focus is on comparing Poisson and Negative Binomial generalized linear models, where the Poisson model provides a natural starting point but relies on the assumption of equal mean and variance. Using the HDHI Admission Data, length of stay is modeled using patient age, type of admission, and selected disease indicators. The models are compared using overdispersion diagnostics, AIC, BIC, diagnostic plots, likelihood ratio testing, Quasi-Poisson checks, grouped cross-validation, and simulated datasets. The aim of the thesis is to determine whether the Poisson model is adequate for modeling hospital length of stay, or whether the Negative Binomial model provides a more suitable framework. The thesis concludes by providing the reader with a guided model choice based on the combined diagnostic and validation results.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Grzegorczyk, M.A. and Krijnen, W.P. |
| Degree programme: | Applied Mathematics |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 28 Jul 2026 06:45 |
| Last Modified: | 28 Jul 2026 06:45 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/37467 |
Actions (login required)
 |
View Item |