Tuesday, June 6, 2017

Data Isn't Just Data

By Hunter Allen

In his book The Information: A History, a Theory, a Flood, James Gleick says that the basis of the universe isn’t matter or energy. It’s data. This is quite a profound and magical statement when you think about it. How we interact with data changes our lives—our cycling lives, but also our life in general, from your smartphone to your internet connection with knowledge from all parts of the world.

Data is just strings of bits. Whether or not it contains information depends on what you do with it and how you learn to interpret it. Interpreting data into useful information is a key skill we all need to improve in all facets of our lives. The more information we have, the more we can understand the role it plays in our lives and how we can become better cyclists and citizens in the world around us.

Using a power meter is one of the main ways we can collect data in our cycling training and racing. However, a power meter can only collect the data. We have to interpret that data into information that can be used to make changes in our cycling. Power meters collect this data in one-second samples in seven different channels, such as speed, cadence, elevation, torque, and even GPS. The key to making use of that data is turning it into information through software analysis in a program like TrainingPeaks WKO+, or education in articles and books.

Each category of data can give us some insight into an aspect of our cycling that we can improve or just learn about for a better experience down the road. Using a power meter on our bikes is really the only meaningful data capture device currently available in our world. (A meaningful data capture device, in my opinion, means that it has the ability to help us make changes in our training; it gives me the information I need to decide whether or not one of my athletes should do a workout, do 8 hill repeats or 12 hill repeats, or train his threshold power or his anaerobic capacity. This information makes my job more precise and efficient.)

There are seven main categories of data:

1. Point-related data (the power, cadence, or HR at a specific moment in a ride). This can help determine if an interval or exercise was executed correctly. It’s the simplest of the data you capture, and it drills down to the minutiae so we can determine if we held just enough watts for the required period of time. I look at point-related data daily with my clients’ files, and it’s something I learn many things from—how many watts an athlete cracked out for the interval, whether or not he paced himself correctly, and even whether he created the watts correctly using the right balance of force and cadence.

2. Warning system data. Data can be used as an early warning system. This data is comprised of many, many smaller data sets, and we need to look at this data over a long period of time. Unfortunately, in order for this warning system to ring the warning bells, your data set needs to include your rest days, your hard days, your races, and all your rides, no matter how easy or hard they are. This is a critical part of the warning system. If you’re missing data because you didn’t use your power meter in a race or because it had to be sent back for repairs, the integrity of the warning system is really compromised. My warning bells can tell me when a client is doing too much training too quickly and when overtraining could occur. Another warning could be a sudden and unexpected drop in your threshold power. While out on a ride doing intervals, you could use your power meter to tell you when to stop doing intervals when your power decreases below optimal levels in creating the right training stress.

3. Detector data. Data can be a detector. How fresh were you when you cracked out your best twenty minutes? When you blew on the big climb, what happened five minutes before it or ten minutes before it? Use post analysis of your data to better understand your failures and successes. When you succeeded, what exactly did you do in order to succeed? When you failed, why did it happen? Was it the tenth hill that crushed you, or was it the violent attacks up the tenth hill that crushed you?

4. Instantaneous data. Data can provide instant feedback. During a workout we continually watch our power meter in order to keep ourselves within required limits for optimal training, and this is where a power meter can help us in pacing. Cycling is a sport of pacing, and you have to pace yourself in a breakaway, in a long road race, in a short criterium, and in a century ride or gran fondo. Pace your effort on the hills. Pace yourself in your nutrition and hydration, as well. These are all key fundamentals to your success as a cyclist, and one of the beauties of using a power meter is that the data is instantaneous. You push down on the pedals and see the number on the screen instantly. There’s no lag time, nothing to wait for or download later. It’s right there, and it happens immediately.

5. Investigative data. Data can help us be detectives. If a problem occurs, we can use the data to help us detect the cause. Sometimes we have to dig deeper into the issue surrounding a success or failure, and reviewing the data may be the way we discover the true underlying cause of our performance. I spend a lot of time being a detective when I analyze an athlete’s data, asking myself questions like, “How many times did he have to attack, and how many watts were in each attack before he was able to get away?” or “As this athlete fatigues, does she choose a bigger gear because she has more natural strength than endurance, or does she just not have enough muscular endurance to begin with?”

6. Explanation dataData can explain why we go faster or slower than normal, but we have to understand what information the data is trying to tell us. We have to translate it. Like James Gleick said, “Data is only a string of bits and has nothing to do with information. The information comes from understanding, and that is our job—to understand it.” Why were your watts lower than yesterday? Is it because you were tired and couldn't physically produce them? Was it because you tried to test up a steep climb when you’re better as a flat time trialist? Was it because you were chasing your arch nemesis and therefore pushed harder than ever to beat him? This type of data is similar to the data we get from investigative data, but explanation data provides a quicker insight into the information you need.

7. Incorrect data or biased data. This kind of data is worse than no data at all. Sometimes we can adjust for incorrect data from our past experiences, but other times we have to throw it away. Incorrect data is easy to identify in most cases, but biased data is much harder to discern. Fortunately our power meters aren’t biased (I hope!), and therefore we rarely have to consider biased data, but our data can often be incorrect, which can pose many problems in analysis.

The data is always clear as a bell to see, but it's not always clear whether or not it explains the problem. We must first prepare the data in order to identify the problem, and this is what turns data into information. To achieve the right interpretation of the data, we need experience and a gift for joining the dots together in one picture (or just good computer software). I do believe you need to have a personal connection to the data and understand this information first for yourself before you can understand it for others. I’ve seen too many coaches trying to coach athletes with a power meter without having ever used a power meter themselves; they have no understanding of what 300 watts or 1,000 watts feels like. This data—this information we capture on a power meter—is unique in that we can associate it with a feeling and learn that sometimes our feelings are incongruent with the data and other times match exactly how it appears.

Experience and a basic knowledge of riding and racing a bicycle are essential. We are creating a harmony between man and machine. We’re trying to optimize what our bodies tell us about how it feels and what the data tells us about how we feel. Relying solely on data is dangerous and doesn't tell the whole picture, but the information we gather from the different categories of data can help us improve as cyclists and citizens of this world of data.

Hunter Allen is a USA Cycling Level 1 coach and former professional cyclist. He is the coauthor of Training and Racing with a Power Meter, He helped develop TrainingPeaks’ WKO+ software, and is the CEO and founder of Peaks Coaching Group. He and his coaches create custom training programs for all levels of athletes. Hunter can be contacted directly through www.PeaksCoachingGroup.com.