SwingDJs

Here's a super-nerdy question for y'all. Imagine that you were going to do post-game performance analysis on a DJ, either yourself or someone else. What statistics would you look at and how would you set up the analysis?

Here's two examples of simple bpm charts:
Frequency distribution of tempo ranges
Time-series of track bpm over the evening

Both of these graphs use information that you can extract from a playlist, so easy to execute if you are a laptop DJ. Similarly, it'd be pretty easy to look at the last time you played any particular song in your set, or how frequently you've used a particular song over your past 12 sets. What else would be interesting/helpful?

I'll leave the question even more wide-open--so imagine that you can collect data on anything at all. If you'd be interested in measuring the number of balboa toss-outs executed over the course of your set, just speak up. Thoughts?

I oten do post-set think-throughs of my sets. I found that when I was doing that regularly my DJing was stronger (I'm kind of needing to pick up my game atm). I'm not all that keen on stats; I'm more of a qualitative analysis chick. The next day (or sometimes that night) I write down what I remember happening on the floor, and hypothesise about why that was happening. It's very useful, but not for everyone - I think it's an ob-con job.
I like it that itunes remembers what time I played which song during the set - it gives me an idea of what people were dancing to at 2am, 9pm, etc.

I used to put charts like these together to disprove the misconception that I played nothing put fast old music. I love playing with numbers, however I rarely have full night sets any more, which are the type of sets that can produce meaningful data. But, I do have a nice big set list from this years ULHS to analyze. I think I'll play with that tonight.

Nevermind... attempting to do real work with Apple applications is a nightmare.

This is going to be a perfect procrastination project for me tomorrow... makes me wish my music collection were actually set up as a proper database. You could sort things by bpm, artist, style/genre, instrumental vs vocals, male vs female vocal, primary instruments, etc.

zipthebird wrote:... so imagine that you can collect data on anything at all. If you'd be interested in measuring the number of balboa toss-outs executed over the course of your set, just speak up. Thoughts?

The first thing I try to do when DJing from iTunes is make a manual playlist record of the evening by going to the main Music list, then sorting by 'Last Played', then selecting all songs played on the night, then making a 'New Playlist From Selection' from the File menu. I give it a name like 'Swing Club Tuesday 18th Nov 2008'. This creates a permanent record in iTunes (although if you happen to play the same song twice in a night, it only records the last time you played it).

Then I look at the overall shape of the setlist in terms of tempo, sound and feel. And I look at individual songs. I try to remember the reaction, in particular the reaction to new songs I played - did they work, or didn't they? And I look at patterns or transitions eg. Slow Song K followed by Medium Song E.

I think I naturally tend to remember what went well rather than what didn't, but sometimes it's almost more valuable to look at what didn't work, and then try to understand why.

Mr Awesomer wrote:I love playing with numbers, however I rarely have full night sets any more, which are the type of sets that can produce meaningful data.

That's probably true. But even if you regularly DJ for an hour or two every week or two weeks, you could look at trends over your last couple outings. Are your sets composed of 50% material that you've played in the past month, or almost entirely of music you've played less than twice ever before?