These extend the ordinary regression framework to situations where the data do not follow a Normal distribution, or where a transformation needs to be applied before a linear model can be fitted. The two most important distributions are the binomial distribution, for discrete proportions, and the Poisson distribution, for counts. Links (i.e. transformations) include log, logit, power, probit, reciprocal and square root.
Generalized linear model
Updated on December 4, 2017