regression
Poisson regression
Model count outcomes or rates.
Requirements
- count outcome
- predictor(s)
- offset for exposure when rates differ
Avoid when
- overdispersion without robust/negative-binomial handling
Formula
log(λ) = Xβ; Y ~ Poisson(λ)
Interpretation
Exponentiated coefficients are incidence-rate ratios.
Worked example
Model a count of hospital visits from exposure and patient characteristics.
Open example datasets