I have written an application that analyses listening patterns in iTunes and generates track ratings automatically. This is particularly useful when using Party Shuffle. The application awards a high score to tracks that are listened to frequently, and penalises tracks that are skipped frequently.
It uses the formula:
rating = (100 * ( (play frequency – lower) / (upper – lower) ) – skips per month * 5
Where upper is the main play frequency + 2 standard deviations, and lower is the mean play frequency – 2 standard deviations
Download AutoRate 1.0 here