Learning Curves chart the model's training and test metrics across various training data sizes.
Use it to identify issues and decide actions like adding or removing features, adding more training data, or modifying parameters.

Reading Tips

Underfitting
Poor training and test metrics.
Add features or modify parameters.

learning curves reading tips illustration learning curves reading tips illustration

Overfitting
Good training metrics but poor test metrics.
Use fewer features or add more training data.

learning curves reading tips illustration learning curves reading tips illustration

Good Fit
Training and test metrics are good and stable.
No changes needed.

learning curves reading tips illustration learning curves reading tips illustration

Improving
Test metric keeps improving.
More training data will likely improve performance.

learning curves reading tips illustration learning curves reading tips illustration