Wednesday, June 29, 2016

Understanding Data - Part 1 of 2,789,003

I got into an interesting debate with my dad yesterday. To qualify, he claims he is not all that tech savvy, but he is a better than average user of computers, phones and the like. Part of the discussion was about too much data.

I don't believe it is possible to have too much data, especially now that we have so many tools at our disposal to make sense of what we capture. I even have a spreadsheet where I try to capture all of the results of my blood tests, before and after cancer. It paints a certain picture.

Now in regards to blood tests, I can potentially see where having a test every week is overkill. I get tested at least 4 times a year, and last year had 6 or 7 tests. Through those tests I was able to detect that I had gone from being hypothyroid to hyperthyroid in early spring, and that I was actually having issues back into November 2015. Data can be good.

Back in the early 1980's we were all so happy when the Avocet computer came out. We could get miles, speed and things like that in a fairly accurate little unit. It was funny how we even advocated taking the battery out on January 1st to start over on our data collection. Now Strava does that for us.

I used to be a premium member on Strava. I recently shut that off. I didn't feel that I was getting much from this tool as I couldn't really work with my data. It was not even equal to the Garmin site where I could break down power a little bit better, see balance, and get peak numbers. I was a bit disappointed with Strava in regards to number crunching. However, I do love comparing times on segments and the like, in reality this is one type of number crunching, and kind of old school. When we didn't have power and heart rate, if you got dropped, you knew you had work to do. If you stayed on, you were doing ok.

I have been using the Watteam PowerBeat, a low cost ($499) power meter that is DIY installation. Easy To Use
Effectively, you are given a tool that locates the sensor on the right of the picture onto the crank arms, and then bond them into place. The sending units on the left send a signal and data to a Garmin or similar computer for calculation and storage of the data.

I have been happy somewhat, since I didn't have a powermeter for a while, and feel that when paired with many things that power can be a valuable tool.  I did have an issue with one sensor, and had to replace it. My other issue that I currently have is that the sensor placement is a little vague. If that sensor it placed a few millimeters too close to the bb, it will not register enough watts, or too close to the pedal and it will detect more flex and register too many watts. How do you really know what the most optimum placement is and get an accurate number?

Simplified Powermeter Theory

A small part of the power debate is getting a number and sticking with it. Most people are happy with that. They get their power number and they can brag or whatever. Here is the reality, why do you even capture data about your training?

I would hope that the goal with using a Garmin, or some type of computer is so that you can analyze the data, and use it to improve. Now here is where the debate begins, what data is important, how much helps, and can you focus on the data to make sense of it and derive value? How exactly do you adjust and improve? What is the best way to do so?

In our scenario, how exactly is your power meter sampling? Remember the meter is digital, and not analog. Based on the processor, it will read from its sensors so many times per second. Next, your computer is recording data, but how often? The default interval on Garmin is 2 seconds. So, your powermeter is sampling potentially 10 or more times per second, it might be sending a data point to the computer less times, and the Garmin is recording every 2 seconds. You might see a potential fallacy building there. But again, you are getting a number from your powermeter to judge performance on. Is it the best number that can be calculated, maybe yes, maybe no. It depends on the time period, and how many times it can derive that accurate number in the time frame.

Realizing this, other manufacturers spec out their own computer, such as SRM or Pioneer. While it is possible to record data from either manufacturer with a Garmin or similar computer, the results will be more accurate using the manufacturer specific computer. Like we were discussing above, the powermeter itself is capable of sending a reading multiple times per second, the sending unit may perform some localized calculations and then send data to the computer, but it is possible to get more data points with this scenario.

Data Points

So, you may hear someone on a ride or a race boasting some power number. Or you hear about watts per kilogram for Tour de France riders. What does all that matter? Well it is a relationship that we are attempting to define, and it all revolves around data points.

We all love big numbers. There is a reality though, big numbers can be nice at the right time. But think about this, I don't want big numbers all the time. Actually I want small numbers most of the time, more specifically I want small watt and heart rate numbers in the middle of a race, and to maintain those small numbers while remaining in contention for the win. At a given time, I want max or near max numbers to be available to win the race at either a climb or sprint point. A realization is that, yes, we must be able to maintain big numbers at the right times.

But big numbers. What if that number is a single data point? Is it good? Yes, it is good, you got the number, maybe one time in 50 miles. It means something. If you did it twice, even better. Maintain that number over a longer period of time (data points), and you are on to something.

But... Is your powermeter capturing your big numbers? This is where you need to record many data points. The more the better. In the case of a Pioneer unit, it attempts to capture a reading every 30 degrees throughout your pedal stroke up to 120rpm. Because it is capturing data that frequently, it can start to do more with that data. It can start to paint a picture of your pedal stroke.  
In red, it shows to varying degrees how much power is being applied, in this case during the down stroke on the left and right hand side. By having significant data points, it can draw this picture accurately and provide something that a single power number cannot.

The goal with training, and any of these tools is to improve your performance. Yes, you can improve your performance tracking a single number at a single point. Yes, you can improve by having an overwhelming amount of data as well. If you take the time to make the data mean something, build a relationship with it. Effectively Strava does that one way, which is time over distance.  We are not yet sophisticated enough to understand power over distance, or to equate with sensitivity power to effort and given results. Yes, there are a few of us that might be able to, but unfortunately the majority of us are fixated with 20 min power, or even 5 second surges. It makes us feel good, and in a way, that did improve your performance.

No comments:

Post a Comment