Depending on the kind of target field, the resulting model type will be either classification (for categorical target field) or regression (for numeric target field). The predictor uses a table of historic records for which the target value is known to train a model.
The model can then be used to predict the target field value for other records for which the target field value is unknown.