Binary Logistic Regression is linear regression with binary dependent variables.

Model for logistic regression:

 

  ,

where

are probabilities of the dependent variable equaling 1,

 are coefficients of the regression model,

is j-th sample of i-th independent variable (factor or future).

 

Regression coefficients are estimated by maximizing log-likelihood function:

 

 ,

where log-likelihood function equals,

(see Logarithms Exponents Sum function).

 

Syntax

 

maximize

 logexp_sum(matrix_1)

 

Parameters

 

matrix_1 is a Matrix of Scenarios:

       

where .

To evaluate probabilities use PSG function Logistic.

Example 1. Estimating probabilities in Optimization problem (see Logistic Regression in Run-File Environment)

 

maximize

 logexp_sum(matrix_1)

Value:

 logistic(matrix_1)

 

Example 2. Estimating probabilities using Calculate problem (see Logistic Regression in Run-File Environment)

 

maximize

 logexp_sum(matrix_1)

 calculate

Point: point_problem_1

Value:

 logistic(matrix_1)

 

 

Example