When working with a regression data set, some conditions may not be met and there may be limitations to running the
 regression model, which makes it difficult for us to use the least squares method. The generalized maximum entropy method
 with regression infrastructure is able to estimate the parameters of the regression model without considering any conditions
 on the probability distribution of errors. The capability of this method in small sample sizes has already been investigated
 and confirmed. When the response variable is a qualitative variable, the logistic regression method is used. In this study,
 we first introduced the generalized maximum entropy method for a logistic regression model. A random sample of bank
 customers was collected and in this study to estimate the model parameters from the binary logistic regression model using
 twogeneralized maximumentropy methods andmaximumlikelihoodanalysis and statistical work was performed and finally
 compared the two methods. Based on the mean square error statistics for predicting customer demand for long-term account
 opening, which was obtained from logistic regression using the generalized maximum entropy and maximum likelihood
 methods, it was found that the generalized maximum entropy estimation method has accurate results.