In our upcoming 4-part blog series, we will discuss how to best utilize biometric analytics in machine learning development, and how sporting good manufactures & professional teams can best use this data to make better decisions.
Biometrics are a great data science tool
At some point, we have all experienced the feeling of picking up the perfect tool for the job. Whether a hammer, a guitar, a golf club, or even a piece of software, there is something magical about a tool that just feels right and works. This synergy between equipment and user is particularly important in the sporting goods industry. Sports apparel, footwear, and equipment are, after all, designed to enable athletes to perform at their highest level. Innovation in manufacturing and engineering over the past decades have provided valuable insight into revolutionary materials and designs for equipment, and now, in the age of data science and artificial intelligence, there are significant opportunities to understand athletes in an equally innovative way.
Collecting & analyzing data isn’t free
Ultimately, a key goal for the sporting goods industry is to get the right equipment into the hands of the right athlete. There are many sources of data that support the stages of this process including design, marketing, distribution and retail. Data in this space range from detailed scientific measurements for a specific athlete to broad consumer metrics across the sector. Each of these data points requires a level of effort, or cost, to collect and provides a commensurate level of detail about a given athlete or group of athletes. In addition, as data become more detailed, they will also tend towards being more specific to an individual or small segment of the population. In other words, we can spend a certain amount of time and money learning every detail about a single athlete, or we can spend that same time and money learning a little about a large group of athletes.
In scientific research, and for many top tier professional athletes, the time and cost of collecting extensive data are feasible and justified. These detailed data points might include biometric analytics data that simply cannot be collected outside of a laboratory setting, such as electromyography, physiological chemistry, or detailed whole-body motion capture videography. These types of data provide a rich description of individual athletes and performances, and they have led to many breakthroughs in the design of footwear, golf clubs, protective equipment, and even sports apparel. However, this type of data collection requires specialized expertise, equipment and often many hours of time devoted to every athlete collected.
At the opposite end of the spectrum, there are significant opportunities from a broad consumer/retail standpoint. Information about consumer demographics and purchasing habits are key pieces of data that inform distribution, marketing and retail level strategies. It helps that these data are often already being collected by manufacturers and retailers, putting this approach easily within reach. On a more detailed level, understanding size profiles for a consumer group can provide a path to optimizing allocation of product and resources to the right retail locations to increase sales, reduce stockouts and improve the bottom line. Although this is a somewhat more ambitious approach, historical data on sizing is often available from purchasing histories, and there are opportunities to infer size profiles from open access data sets and general knowledge of demographics and body sizing.
Looking at these two extremes of highly detailed athlete-specific data versus general knowledge of the consumer population, there seems to be a large gap in between. Can we begin to bridge this biometric analytics gap? We think that this is possible, and that there are significant opportunities to be explored and leveraged within this gap. There is no better time to dive into these opportunities, given the deluge of data that is becoming available through online communication and technological advancements; these data provide a key resource to pursue these opportunities.
Real-world use case | Amateur Running
Consider participation in amateur running events, which has increased dramatically over the last couple decades. Many of these events are heavily documented online through the event website, social media postings, and even published photos and videos. These data provide a treasure trove of information about basic demographics for these athletes, their geographic location, and their level of performance. This gives a footwear manufacturer, for example, unprecedented visibility into the characteristics of their consumers. Not only that, but it also paints a picture of the athletes that AREN’T currently customers, but could become customers in the future.
Recent technological advancement in video capture, online storage, and computer vision is also set to have a disruptive effect on how we can learn about athletes and sports performance. Whereas in the past, only top tier athletic events would receive widespread coverage, it is practically commonplace nowadays to film amateur events and post videos of the athletes in action. Using the latest in computer vision and artificial intelligence techniques, we can measure or infer a number of details about athletes from video, ranging from the simple (height, weight, body shape) to the complex (kinematics and physiology). Coupled with the miniaturization of many sophisticated measurement systems (cameras, GPS, accelerometers), these hardware and computational technologies provide a step towards athlete-specific analysis from accessible data sets.
Fulfilling the promise
The only way to unlock the full potential of this new frontier in biometric analytics is to apply tools from data science and artificial intelligence to sort, identify, and extract the relevant features from the massive oceans of data being generated every day. Doing this will close the gap between the laboratory and the field, and will serve the ultimate purpose of putting the right piece of sports equipment in the hands of the right athlete at the right time.
