HR windows - why the lumps in the data?

[carpetrunner]

I've been looking at some HR analysis and I'm looking for answers.

The data below is from Blayney2Bathurst 110km in 3h11m - so I was pushing it all the way.
The course undulates a little with a reasonable hill at the 85km mark.
http://app.strava.com/activities/49869120

I was thinking that maybe I could extract some target zone HR data from over 3h of 3sec samples. If the average power for a flat out time trial for 1h gives a reasonable approximation to power at LT, then the highest result of a 1h average should be my average HR at LT... 88% of HR(max) ok so far.
Then I was thinking I could also calculate a few more time windows to look at say my average for 15min, 30min, 90min... and there were a couple of time intervals that went backwards in HR :crazy:



I'm guessing mr Nyquist is playing tricks with some underlying frequency - but what would be causing this?

Is this kind of data normal?

- carpetrunner

 

[cha_cha_]

Ok, I'll play:

1) hr data's correlation to power output is poor at best. It improves as the interval duration increases, but it its never, ever good.

2) it is a big assumption to say that this ride is your best effort across all time durations. Normally you can do a best effort for a short duration but that maximal effort will preclude you from doing any longer maximal efforts and vice versa. Good quality critical power curves take weeks or months of riding including genuine testing to properly develop.

3) when your average number vs time goes up while travelling left to right across the graph, then that means that somewhere within that effort is a period of sub-maximal effort that had drawn the average down followed by a higher period to bring it back up. E.g. an interval of a min @500w followed by a min rest @100w followed by another minute@500w week have a higher avg power at 3mins than at 2mins. This is all well and good for power data, but with heart rate it's hard to say exactly what it means - maybe you drank lots of water or had a gu or rode in the shade for a few minutes or something which was enough to reduce your heart rate for a while then you started busting a nut again. Idk...

What you're trying to do here is analyse data in a way that the data doesn't really lend itself to be analysed. Ultimately, it can't really tell you a great deal beyond perhaps some highly subjective target heart rates for races or training intervals but you'd probably be better off doing those based on perceived effort anyway.

It does however sound to me like you're fully nerding out on your data so seriously, do yourself a favour and just go and buy yourself a cheap, 2nd hand srm/quarq/powertap, collect more valuable data and analyse the crap out of it. I saw a used srm go for under $400 last week so they're definitely getting cheaper...

 

[carpetrunner]

It does however sound to me like you're fully nerding out on your data so seriously, do yourself a favour and just go and buy yourself a cheap, 2nd hand srm/quarq/powertap, collect more valuable data and analyse the crap out of it.

Well picked Cha_Cha - I have already purchased a powertap and I'm reading up on how to nerdout on mountains of great data.

I'll post more silly graphs later.

- carpetrunner