40 Algorithms Every Programmer Should Know

Understanding overfitting

If a machine learning model performs great in a development environment but degrades noticeably in a production environment, we say the model is overfitted. This means the trained model too closely follows the training dataset. It is an indication there are too many details in the rules created by the model. The trade-off between model variance and bias best captures the idea. Let's look at these concepts one by one.