Abstract
In this paper, a general model is proposed to extend generalized linear models to non-monotonic link functions. In order to determine the best model, different link function families are analysed, and through AIC, the best model is chosen. Moreover, using asymptotic properties of maximum likelihood estimates are calculated with confidence intervals and hypothesis tests for each of the parameters. Lastly, one example of regression extension is proposed for the Bernoulli distribution applied to in vitro germination of Leucocoryne coquimbensis seeds.
Original language | English |
---|---|
Pages (from-to) | 448-457 |
Number of pages | 10 |
Journal | European Journal of Scientific Research |
Volume | 61 |
Issue number | 3 |
State | Published - Oct 2011 |
Keywords
- AIC
- Bernoulli distribution
- Generalized linear model
- Link function