This method estimates coefficients of the regression model, in which the dependent variable may take the values 0 or 1. The following models are provided in the method:
Logit model presumes that occasional errors of a model are distributed in accordance with the law:
Probit model presumes that occasional errors of a model are distributed in accordance with the law:
Gompit model presumes that occasional errors of a model are distributed in accordance with the law:
Consider the example of linear regression model:
yt = 1 - F(-x'tβ) + εt
t = 1, …, n
Where:
t. Observation number.
β = (β1, β2, …, βk)'. A set of unknown parameters (coefficients).
εt. Random error E(εt) = 0.
Yt. May take the value 0 or 1.
See also:
Library of Methods and Models | Modeling Container: Binary Choice Model (Maximum-Likelihood Method Estimation) | ISmBinaryModel