Part 2: I analyzed the chords of 1300 popular songs for patterns.

&#62; What stands out here, is that IV → I (F to C) is not only normal, it actually shows up just as often as V → I. This is surprising (at least to a classically trained person).<p>That's not surprising at all, and I don't believe for a second that a classically trained person would think so either. V -&#62; I is a common movement for resolution, so is IV -&#62; I as are several others. While his data is interesting, I find the analysis weak, he's acting surprised at perfectly known and normal things, even to someone classically trained IMHO. IV -&#62; I isn't "breaking the rules".

Relevant:<p><a href="https://www.youtube.com/watch?v=5pidokakU4I" rel="nofollow">https://www.youtube.com/watch?v=5pidokakU4I</a>

For an interesting visual way to map common chord progressions, also see <a href="http://chordmaps.com/mapC.htm" rel="nofollow">http://chordmaps.com/mapC.htm</a> for in the key of C. Steve also made a generic one: <a href="http://chordmaps.com/genmap.htm" rel="nofollow">http://chordmaps.com/genmap.htm</a>.<p>I have this as a reference on the piano, and it is nice to experiment with several routes to return to the root chord (nicely positioned at the bottom of the page).

This is much better than the first part.

It's nice to see the data, but I must say I V VI IV was not surprising at all. I'm sure it's slightly more 'emotional' cousin VI IV I V (think apologize) is high up on the list.

Please train a Conditional Random Field on this data. (A hidden markov model would also be interesting, but risks going out of key more easily).