Friday, April 1, 2011

April 1st - First analysis

So one might ask, how does my week 1 performance predict my week 2 performance?   From a correlation standpoint it's actually pretty good, as one would expect.  The elites are still elites and less elite (such as myself) are still not elite.  If you haven't done wk2's exercise quite yet, you can use the plot to plan some sort of pace.  Of course you should try as hard as you can, but the plot might allow you to set some goals from the outset.

Correlation is around 0.7, for people wanting to know.
Part of doing this analysis was wondering if my meager weight might have something to do with my pitful performance in wk1.  While we can't all be Chris Spealler, I figured being a buck40 didn't do me any favors.  Not so.  There's almost zero relationship between weight and wk1 performance.  What correlation ther is most affected much heavier competitors.  Probably those double unders.

Correlation: -0.05

The weight issue, though, in week 2 becomes interesting.  While the data is all over the place, it looks like performance increases from low to about 180, but performance takes a hit after that.  Maybe the box jumps get harder for heavier folks?

Because of the larger penalty as things get heavier, correlation is -0.2.  Clearly the relationship is more complicated if you look at the plot though.


  1. This is super interesting - keep it up!

  2. Weird, when I did the correlation, I got .49. Did you do any preprocessing of the data to cull outliers or anything?

  3. Interesting how there are dense regions of data leading up to complete rounds, esp 5, 6, & 7. I would guess this is evidence of the mental challenge to complete a full round, and the difficulty of starting a new round in the final seconds

  4. @Jeff -- The density of the data leading up to complete rounds in Week 1 might also logically be due to the fact that for most people, the double-unders are fairly easy, whereas squeezing out a few extra snatch reps in the last round is more challenging. The snatches were the rate-determining step, if you will.

    Very interesting data!

  5. I did a histogram of the reps in both WOD 1 and WOD 2 and there are HUGE peaks at "complete round" and "finish double unders". I have a sneaking suspicion that there are lots of people "rounding up". Think about it this way: there should be about the same number of people getting 4+30+12, 4+30+13, and 4+30+14 reps, as that's only a slight increase in speed. I would expect a very small jump up to 5+0+0 because of the motivation to finish a round, then a dip at 5+1+0 because of the mental aspect of starting a new round, but what I see are huge peak (>5x) going at the n+0+0 and n+30+0 reps. I have a bunch of data on facebook (just made it easier to share). It should be public, so go take a look:

    Steal the plots if you want, I don't care.

  6. @miked - thanks for the heads up on the correlation coef. I'm reporting 'R', while you're reporting R^2, which is honestly the more correct value.

    @Jeff - I think you're spot on with your analysis. Interestingly, the density tends to go away at the lower rounds, presumably because those athletes tend to have trouble with the double unders in the first place.