To calculate the prediction, invert the link function for the model. The inverse functions are in this table.
where is from the training data only when there is a test data set for validation.
The confidence limits use the Wald approximation method. This is the formula for a 100(1 − α)% two-sided confidence interval: