AI & DataIntermediate
Overfitting
PronunciationOH-ver-fit-ing
Definition
A modeling error that occurs when a machine learning model learns the training data too well, including its noise and outliers, causing it to perform poorly on new, unseen data.
Where you hear it
In machine learning pipeline discussions, model training evaluations, and data science code reviews.
Examples
The model shows high accuracy on the training set, but its performance drops significantly during testing due to overfitting.
We need to add regularization techniques to prevent the neural network from overfitting.
Common mistake
Believing that achieving a near-zero error rate on training data means the model is ready for production.