Practical Predictive Analytics

Area under the curve

Area under the curve (AUC) is another popular measure to assess the goodness of your model. Historically, this measurement was developed during World War II. The original terminology was Receiver Operating Characteristic (ROC) and its original purpose was to determine whether or not a blip on the radar screen was an enemy ship or just random noise.

One of the things that the AUC tells us is the ratio of the true positives to the false positives. The AUC is determined via a mathematic formula, which will be a number between 0 and 1. An AUC of 0.5 is considered a random classification. Look for points that hover near the upper-left quadrant. This would be the area where advantageous conditions converge: high true positives ...