A model that fits its training data too closely and generalises poorly shows:

1. It performs well on known data only. 2. New data gives high variance. 3. That problem is overfitting. _Source: OpenStax Principles of Data Science, Chapter 7, Neural Networks._