In machine learning, what is overfitting?

A$)$ When the model performs well on training data but poorly on new data

B$)$ When the model performs well on new data but poorly on training data

C$)$ When the model has too few parameters

D$)$ When the model has just the right number of parameters