We can plot the ROC Curve (TP against FP as we vary threshold).
Dotted line is performance of random classifier (on average). Area under curve is related to probability that the classifier will correctly classify a randomly chosen example. The closer we are to the top left, the better the model is.