Threshold Explorer
Move a classification threshold and watch individual predictions, confusion counts, ROC and precision–recall evidence change together.
What to observe while you experiment
Thresholding turns continuous model scores into decisions. Moving the threshold changes which observations are predicted positive, which changes the confusion matrix and traces points along ROC and precision–recall curves.
Experiment deliberately
Choose a cost preference such as “missing positives is expensive.” Predict which direction the threshold should move and verify the resulting recall/false-positive trade-off.
Thresholds change decisions, not model scores. The same scored cases are reused while the cut-off moves from 0 to 1.
Ready.
Scores and decision threshold
Cases do not move. The vertical threshold moves and predictions flip as it crosses their scores.