Data That Matters, Part Two

posted Feb 28, 2010, 5:49 PM by Donald Vescio   [ updated Dec 30, 2011, 10:05 AM ]


In the last post, we tried to distinguish between data that is simply interesting (or depressing!) from data that we can use to track and predict performance. This month, I'd like to suggest a couple of strategies for using data to assess specific performances and to predict how you might perform in the future. 

As a cyclist, I'm very interested in my power output as it's a fairly objective measure of the work that I can perform while riding.  But having a power meter and tracking average power for a given training session really doesn't offer much in the way of useful data; rather, for the data to be useful, it needs to be placed in a broader context. 

For instance, suppose I scheduled a workout based on six VO2 Max intervals of five minute duration each.  Because heart rate will drift over time and can be impacted by environmental conditions (excessive heat or cold, etc.), it isn't a consistent measure of performance during each interval.  Similarly, average speed won't work well, either, as terrain and wind are variables that are difficult to factor in during a post workout session analysis.

Power, on the other hand, is remarkably constant--work (in this case wattage) can be accurately measured independent of environmental variables.  So I do my six intervals and download my data later that night.  What I'm looking for is not so much my peak power during each of the intervals, but rather my average power across each of the efforts.  For example, suppose that I captured the following averages for my six interval workout:

Interval 1 = AVG 381

Interval 2 = AVG 403

Interval 3 = AVG 401

Interval 4 = AVG 398

Interval 5 = AVG 400

Interval 6 = AVG 329

What is important to me is how the data trends across all six intervals.  Looking at the data, it would appear that I could have gone a bit harder on the first effort and that my sixth effort was subpar and yielded little benefit.  If I were to schedule this session again for six intervals, I most likely would slightly reduce my effort during intervals two through five so that I could complete a quality sixth interval.  In terms of my intent for the session, the data above suggested that I did not achieve my goal of completing six quality intervals.

Data Over Time

The analysis above is remarkably simple and in actuality I would also attempt to identify additional patterns, such as whether my power trended upward or downward over the time that it took to complete the interval.  A similar sort of analysis can be done with heart rate data for runners and swimmers, though again environmental factors should be considered.

What really is useful are the larger patterns that emerge over time in your data collection.  There are a number of algorithms that can be used to quantify the amount of cumulative stress your body records as you complete your days and weeks of training.  We all have experienced instances when we are exhausted come race day; similarly we all wonder whether we are doing to much (or too little) training in the weeks leading up to an important event.  This is where your data (and a good laptop!) comes in--a good software package will automate most of these calculations, leaving you with the opportunity to focus more on your actual training.

TRIMPS ( which is a time in zones multiplied by a weighing factor) has been used my many different types of athletes since the late '70s; more recently, CyclingPeaks Software has introduced TSS (total stress score) that enables multisport athletes to track cumulative stress of their activities, and PhysFarm's TriUtilities' GOVSS extrapolates power-based efforts into swimming and running workouts.  Polar's Precision Software is an easy to use application for those who track heart rate data.

The example below was taken from Polar's software and tracks one week of recovery, followed by three weeks of increasing training intensity:

Looking simply at the vertical axis of the graph, I can immediately see that my actual training activity matched what I had planned in advance.  In the example below, my taper prior to FIRMman RI also is evident.  If there wasn't a drop in training stress prior to the large spike associated with FIRMman, then I would have known that I had not prepared sensibly for such an intense race.


What to Take Away?

In short, data collection and analysis can be a powerful tool in optimizing your workouts, in assessing prior performance, and predicting future performance. Keep in mind the following:

  • Collect data that can vary over time

  • Collect data that accurately measures your effort (such as power) and the impact that the effort has on your body (heart rate)

  • Look for micropatterns within individual workout sessions

  • Look for macropatterns that emerge from your data over time

  • Consider adopting one of the many excellent software packages that will help you analyze and compare your data

Go to part three