Researchers Predict Hit Potential Dance Tracks
Researchers from the University of Antwerp have developed a data mining model which recognizes potential dance hits. The university states this in a press release.
Using new technologies, they created a prediction model which very accurately calculates the chance of a track reaching the top 10 of the dance charts. Researchers Dorien Herremans, professor Kenneth Sörensen and professor David Martens of the Department Engineering Management (Faculty Applied Economical Sciences) created a database of dance hits of the period 1985 until 2013, with a number of basic characteristics for every hit, such as timbre and the number of beats per minute (BPM).
They also mapped more advance characteristics, such as the evolution of the number of BPM throughout the track. "During the analysis of many dance hits, we noticed a number of clear trends, such as the increase in loudness, the clarity and the speed of dance hits through the years."
They scanned the database and looked for common characteristics. "Using data mining techniques, we drew up a prediction model which very accurately calculates the probability of a track reaching the top 10 of the dance charts, versus the probability of it ending up in the top 30-40." The developed model recognized a stunning 83% of the tracks correctly as being a hit.
"In our free online application, a user can personally upload an mp3 and immediately see if his/her dance track has a chance of being listed in the top 10". Those who want to check the hit potential of a dance track, can do so here. Practical for your next event!01-12-2014