A Hit Prediction Idea

June 3, 2013


I’ve got a hit prediction idea. What if we were to statistically represent the redundancy pattern of a song? Determine how uncertainty during the song’s course plays a role in its popularity. Refereeing to Shannon’s work on information theory, would a more predictable song determine a higher popularity in terms that fewer (sic-) information was relay in its message entirety?

That is to say, what is music’s entropy and its correlation to popularity?

I would imagine though that each song is a ‘language’ of itself, representing the statistical patterns of the artist’s predicament at that moment of composition. Then again, can one susceptibly find the ‘bottom’ of a song? That, I think, would in effect render it dull. It is the fact that a piece of art, it be storytelling or a piece of music, is intermingled with redundancy and apparent randomness which renders it interesting. Extremes of patternless i.e. random or pure redundancy are a bore.