Algorithms, models and training concepts.
Overfitting happens when a machine learning model learns the training data too closely, including its noise and quirks, so it fails to perform well on new, unseen data.