Dave Krenik

Goodbye Stopwatch; Hello Sapphire Now

Blog Post created by Dave Krenik Employee on Jun 4, 2018

There was a time when all the data you needed to effectively train for endurance events, like cycling and running, was a stopwatch with a second hand.  Now power files from the top pros are (largely) kept secret and those of competitors from local amateurs to world champions are dissected by watts, cadence, heart rate, elevation, and more.  This data can be used to answer, or at least give some insight into, questions like:

  • Why did rider “x” lose contact with their group on the last hill?
  • What are my “limiters”?  Or, what part of my fitness is holding me back in my key events?
  • What are the most effective intensities and durations to be training at?


Big Data and Analytics have become such an integral part of the sport that top teams like Team Sky have a Chief Data Scientist. The Chief Data Scientist’s role on teams like Sky is, in part, to provide insights to guide equipment selection and racing strategy.  Do the aerodynamic advantages of deep carbon fiber rims outweigh their weight disadvantage? Knowing a cyclist only has so many kJ of energy to expend in a ride, where are the most effective parts of the course to apply them?


WKO4.pngWKO4 is an “analytics engine” of sorts for cyclists and runners training and racing with power meters.  Tim Cusick is the Product Lead for WKO4 and coach for several world, national, and state champions in cycling.  Tim also comes to cycling with a background in big data, analytics, and machine learning.  Tim was kind enough field a few questions about big data, analytics and how they’re affecting endurance sports.


Dave Krenik:  Tim, would you inform the readers on what WKO4 is and how it leverages analytics to help athletes?

Tim Cusick:  WKO4 is a robust analytical engine specifically designed for endurance athletes. It was devised to evolve the paradigm of performance data analysis (the process of breaking complex data into smaller parts in order to gain a better understanding of it) into true performance analytics (the discovery, interpretation, and communication of meaningful patterns in data).  Analytics is multidisciplinary and uses mathematics, statistics, models, and descriptive techniques, but at the time, the existing programs out there lacked most of these elements, so we had to start from scratch. First, we worked with Dr. Andy Coggan to develop a highly accurate model of the human power-duration relationship. Once we had an accurate model, we needed to build a system that would allow for a multidisciplinary analysis with the ability to add math and statistics.  WKO4 is the realization of these goals.


How does it leverage analytics to help athletes? Well, that answer has three parts.  First, the use of advanced analytics allows historical performance data to be diagnostic.  One of the hardest things for a coach or self-coached athlete to do is to truly understand any particular athlete’s needs.  WKO4’s analytical capability provides a 360-degree view of an athlete from both a performance and physiological standpoint, which significantly improves the ability to diagnosis that athlete’s needs, thereby creating more effective training strategies.


Second, WKO4 gives us the ability to track training efficacy by increasing the timeliness and amount of analytics of the dose-response relationship of training stimuli and adaptation. This enables faster insights and training adjustments to ensure results.


Finally, analytics solves problems. All athletes train with a plan until something goes wrong or does not work. WKO4’s analytics supply deep insights into athletes’ performing physiology, and these insights lead to better solutions during crisis or change in training.


DK:  Does WKO4 use any of the of the available big data/analytics technologies like Hadoop?  Spark? Why/why not?

TC:  Well, before answering that, let’s clarify an important difference. Hadoop is a distributed file system, which isn’t applicable to WKO4; WKO4 is a desktop application and does not have that much data to require such a thing. Spark is simply a computing engine for Hadoop data. So to answer your question, we used standard big data and machine learning to develop our models, and then we took those models and implemented them in WKO4. Users can then apply our models to their own data to gain deeper insight.


DK:  Where do you see analytics and/or machine learning taking endurance athletics?  Catching cheaters?  Automated training programs?

TC:  I think the future is really in the evolution of analytics. If we look at how big data has led the way, we see a three-step evolution in data analytics: descriptive to predictive to prescriptive. What does this mean? Descriptive analytics can be called reports, and their function is to gain insight(s) from historical data with reporting, scorecards, and charts. Simply put, you review the data in well-compiled reports, find patterns on your own, and make decisions. Predictive analytics is the analysis of a variable (or set of variables) that can be measured for an individual to predict future behavior, or, more simply put, a solution. I think WKO4 was on the forefront of this evolution, moving from data analysis in well-formatted reports to using power analytics to create solutions. Our future goals are to continue to push the envelope and develop the third step—prescriptive analytics. This involves using predictive analytics to recommend decisions using optimization and target results, thus evolving from descriptive to predictive to prescriptive. If we think about it from the standpoint of what an athlete gets from his/her data, the evolution is from reports to solutions to decisions, which is an exciting benefit for athletes.


To address the specific question about catching cheaters, I don’t believe that is on the horizon, as you need to 100% sure of the quality of the data coming in, which is the core challenge. An athlete with a faulty data-gathering device, power meter, or heart rate monitor might be labeled a cheater when in reality it was the device that malfunctioned.


DK:  How might all this impact the coaching profession?

TC:  I have a simple answer here. Give a good coach an advanced tool, and you make a better coach. Give a poor coach an advanced tool, and you create confusion and misunderstanding.  The point here is that as data evolves, so must coaches. These advanced tools can revolutionize the coaching business by improving your athletes’ performance, but only if you invest in the learning and use the data to remove bias.


Thanks Tim – much appreciated.


On the subject of analytics, I should point out that SAP’s big end-user and Partner event – SapphireNOW– is this week (June 4-7) in Orlando, FL. Hitachi Vantara (including yours truly) will be present at the event.  During Sapphire we will be showcasing Hitachi’s great experience and expertise in helping customers make the most of their SAP investment for greater than 23 years. In particular we will be showcasing Hitachi Vantara’s integration with SAP on a Video Intelligence for Smart Cities demo.  We will also be demonstrating the new, enhanced, feature/functionality of Hitachi’s Storage Adapter for the SAP HANA Cockpit.  This adapter simplifies management and greatly streamlines processes for Admins working with SAP HANA.


SAPPDI.pngWe will also be showcasing new certifications for SAP’s Appliance and Tailored Datacenter Integration (TDI) programs.  Whether looking to leverage the management simplicity and performance of the Appliance model or wanting to leverage existing resources in the datacenter (TDI), Hitachi Vantara has certified offerings to help meet your business goals and grow with your business.  Lastly, we will be announcing a reference architecture that leverages the data integration capabilities of Pentaho software.  Data comes in several types:  structured, unstructured, and semi-structured. Pentaho Data Integration helps pull all your data sources (SAP HANA transactional databases & data warehouses, Hadoop, and NoSQL databases (like MongoDB)) together and realize the efficiencies of tiering data based on usage and gaining actionable insights from data with real-time access.


Big Data, Machine Learning, and Analytics already have made their way into our everyday lives. Addressing questions like “how hard should I go and for how long?” is only the beginning.  Cookie-cutter training plans will soon come to an end – freeing coaches to perform more value-added work with their clients.


Be sure to come see Hitachi at SAP’s Sapphire Now Conference (booth #1307) – time to ride…