Related Experiment Videos
A large class of models derived from generalized linear models
1Department of Mathematics, Imperial College, London, U.K. j.nelder@ma.ic.ac.uk
Statistics in Medicine
|January 9, 1999
Summary
This study explores five extensions to generalized linear models, including generalized additive models and hierarchical models. These flexible approaches allow for complex data structures and improved statistical analysis.
Area of Science:
- Statistics
- Statistical Modeling
Background:
- Generalized linear models (GLMs) provide a flexible framework for statistical analysis.
- Extending GLMs is crucial for handling more complex data structures and relationships.
Purpose of the Study:
- To describe five key extensions of generalized linear models.
- To highlight the versatility and combinability of these extensions for advanced modeling.
Main Methods:
- Generalized additive models (GAMs)
- Quasi-likelihood methods
- Joint modeling of mean and dispersion
- Hierarchical generalized linear models (HGLMs)
- Longitudinal models for correlated responses
Main Results:
- Five distinct extensions to generalized linear models are presented.
- These extensions are largely independent and can be combined.
- A further extension to dynamic models is outlined.
Conclusions:
- The described extensions significantly broaden the applicability of generalized linear models.
- The modular nature of these extensions facilitates the creation of sophisticated statistical models.
- Future work can explore dynamic forms of these advanced models.