how to dj, really

A while back, I posted a request for DJs to tell me how they picked their next track. The exact question was: “How do you pick what track to play next?”

I got 20 responses, covering 1959 words, ranging from one sentence to full essays. Then, being me, I made some tag clouds (using tagcrowd.com), and did some n-gram analysis in Python.

So here’s the top 25 words:

It is interesting to me that TRACK has twice the frequency of CROWD.

Top 50:

Now we get things like BUILD, BPM, DANCEFLOOR, and DRUM

Top 100:

Down at the bottom: SLOT, BASS, ORDER, FUN, TEND.

After that, I did some analysis of n-grams. A “gram” is just 1 or more words in a row. “more words” is a 2-gram from the previous sentence. Tag clouds do 1-gram analysis, so I started with 2-grams. The top twenty 2-grams (and their numbers of occurences) are:

“try to”, 8
“on the”, 7
“the track”, 6
“tracks that”, 8
“want to”, 5
“so i”, 5
“of the”, 5
“i can”, 5
“i have”, 5
“a track”, 5
“the crowd”, 5
“they are”, 5
“i will”, 5
“i want”, 5
“the energy”, 4
“with the”, 4
“have something”, 5
“work with”, 5
“if i”, 4
(I’ve bolded the ones that are most interesting to me.)

3-grams got pretty useless pretty fast:
“words separated by”, 2
“a sort of”, 2
“i want to”, 2
“from 70 through”, 1
(…etc. Nothing too useful there.)

There were no 4-grams that appeared more than once. Likewise with 6, 7, 8, 9 and 10-grams. HOWEVER. When I ran 5-grams, a key line jumped out a bit:

“have something to work with”, 3
I said when I started analyzing DJ sets based on musical data that most DJ “just play what works” – I’m pleased to see the same verb show up in a text analysis.

This is nothing like a proper study, but it is all I have time for, for now, and I think it’s interesting just the same. Many thanks to James, Stephanie, Nathan, Emmett, Dustin, Emmanuel, Derek, Drew, Alan, Jesse, Christoph, Doug, Tim, David, Jared, Jordie, Quinn, JP, Laughlin, Teagan, Will, and Mike.