Overfitting refers to the situation when a model does a better job of predicting the dependent variable for data that wasn’t used to create the model, than it does of predicting the dependent variable for the data that was used to create the model.