Swimming Australia (SA) has been helping prepare athletes for the Tokyo Olympics, tapping into the power of analytics to help swimmers reach their full potential.
As the governing body for competitive swimming in Australia, SA has a lot of data. Around a year ago, the organisation turned to Amazon Web Services (AWS), hoping to pull all of its data together and end its reliance on Excel.
SA developed a data lake, fittingly known as Atlantis. It comprises two main sets of data -- competition data and training data.
"We're training the athletes to be able to compete, so we want that data then to be able to talk to each other," SA performance solutions manager Jess Corones said.
With the help of AWS services, the organisation has focused on three projects. The first is known as the relay model.
"When we travel, [we] provide the coaches with as much data around the relays as we can, so that the coaches can make really informed decisions about who they're going to swim in the relays, both from the heats to the finals, and what changes they might want to make between [each]," Corones said.
"We want to know what all the other countries are doing -- normally, it's days and weeks and hours of scraping the internet, looking for international results, to try and understand and see where their form is.
Not only is that process time-consuming, but the data was also being recorded onto a spreadsheet.
"There's potential for error, when you're up late at 2am, entering data off of one Excel page into another, there's huge room for error," she said.
Now, it's fed into AWS and returned back to SA in a structured manner "at the click of a button".
See also: How wearable sensors helped the US Olympic team win 121 medals at Rio (TechRepublic)
Another project is the pathways program. It starts with an algorithm developed by the University of Sydney for SA's junior swimming stream.
"In junior swimming … they swim in age brackets. So if you look at a 13-year old's race, a 13-year old that was born in January has to race a 13-year old that was born in December, and as far as maturation and development, those athletes are two very different athletes. There's technically a years' worth of growth and development, but they still have to get in and race," she explained.
Hoping to avoid higher dropout rates among the younger of the age group, Corones said the algorithm became somewhat of an age correction model.
"It looks at the athlete's performance, looks at their date of birth, and it's a little bit like a golf handicap," she said. "So at the end of the race, although they all race in the same race and they will finish one, two, three, four, five, etc, when we run the algorithm over the race, it will then make an adjustment to say, 'Based on the age of this athlete, maturation, this is potentially where the race would have evened up a little bit'."
Lastly, SA has been working on performance benchmarking, breaking down a race into parts centred on starts, turns, and finishes, as some examples, and determining where an individual would rank in each of the metrics compared to international swimmers.
"So trying to understand what it takes to win … giving them the information of this is where the international competition is, this is where the podium athletes are, this is where finalists are, and giving them goals that they're able to hit," she said. "Being able to put that in the coach's hands on the pool deck is a huge, huge benefit for them because that's where we get most of the questions."
SA only did the handover with AWS in January, and Corones said the possibilities are still yet to be realised.
"I would still say it's really in its infancy because the potential of what we can do with the data lake is immense. We're only just starting to tap the surface of what we can do with it, and gradually exposing the coaches," she said. "We're using the data from it for the first time at the Tokyo Olympics, which is really exciting."