Desktop Survival Guide
by Graham Williams
An alternative to the Logistic Regression, for a target variable having a binomial distribution, is the Probit Regression. Whereas the Logistic Regression maps the target using the logit link function, the probit link function is the inverse cumulative distribution function.
Probit Regression is also often used (particularly in the social sciences) to model a continuous outcome between 0 and 1, for example when the target variable records the proportions of a population at different levels.