Logistic Regression Convergence Options

You can set the convergence parameters for logistic regression model estimation.

Maximum iterations. Specify the maximum number of iterations for estimating the model.

Maximum step-halving. Step-halving is a technique used by logistic regression to deal with complexities in the estimation process. Under normal circumstances, you should use the default setting.

Log-likelihood convergence. Iterations stop if the relative change in the log-likelihood is less than this value. The criterion is not used if the value is 0.

Parameter convergence. Iterations stop if the absolute change or relative change in the parameter estimates is less than this value. The criterion is not used if the value is 0.

Delta (Multinomial models only). You can specify a value between 0 and 1 to be added to each empty cell (combination of input field and output field values). This can help the estimation algorithm deal with data where there are many possible combinations of field values relative to the number of records in the data. The default is 0.