Here represents the classification threshold of your binary classifier, e.g. Likelihood Ratio.
Note When theta is high (at the bottom-left of the curve):
- The classifier requires stronger evidence to predict the positive class
- You get fewer false positives but also fewer true positives
As theta decreases (moving toward the top-right of the curve):
- The classifier becomes more lenient in predicting the positive class
- You get more true positives but also more false positives
- This results in higher recall but lower precision