ZDNet Interview- Kevin Logan talks Predictive Analytics using Rapid Miner
Article Published on ZDNet.com By Heather Clancy for Small Business Matters |
The secret behind any compelling wearables app, whether it’s for personal or professional use, lies in how quickly the data being collected can be interpreted. Unleashing those secrets is the focus of startup Body Biolytics, founded by an artificial intelligence expert.
“We are not making a wearable device, we are strictly into software and what we do with the data once it is off the device,” said Body Biolytics CEO and founder Kevin Logan. The company was started as a spinout of Logan’s other business, MACSEA, which uses AI and neural networks to monitor the “health” of equipment and systems on ships (including more than 40 U.S. Navy vessels).
Wearable apps present an enormous big data challenge, Logan said. Pretty much any gadget you can think of can collect hundreds of data points each second. Correlating that information will be a very valuable service, and is at the heart of Body Biolytics’ future business model.
To experiment with potential services and analytics reporting combinations, Body Biolytics is using RapidMiner. “I needed to crunch the data quickly,” Logan said. “Up to this point, what we used was a desktop tool. The reason I explored the platform is I wanted to figure out a way to cookie-cut and move this to a cloud service.”
The net result was that Logan was able to prove out ideas (and discard some) far more quickly than otherwise. This was one factor that helped the company win an innovation challenge by Under Armour, seeking services and applications that work with its Armour39 fitness monitoring system. After making the initial cut, the company had just a few months to get prototypes up and running.
What exactly will these services do? The company’s launch point will be the sports and fitness space, where it hopes to help athletes use information about their heart rate, blood pressure and other physiological data to establish baselines that can be used to track improvements (far beyond just counting steps). It’s also looking at supporting applications for eldercare, according to its Web site.
Ingo Mierswa, CEO and co-founder of RapidMiner, said his company’s predictive analytics services help “empower non-experts” who need better insights into business challenges such as customer churn rates. But rather than telling your team why someone switched to your competitor after it happens, RapidMiner can help identify the root causes and help predict when someone is likely to become an ex-customer before it happens.
Mierswa is “super excited” about the potential for services such as those being dreamed of at Body Biolytics. “Imagine connecting on an emotional level with body data,” he said.
Although it sounds sort of Big Brother-ish, potential applications could include helping predict the mood or health of employees, particularly those who might be working in field service positions where safety is an issue and an alert might be useful if someone becomes distracted or tired. Many of these same ideas can be applied to machine-to-machine information. “There’s very little difference to predicting whether a machine is going to break, versus when a human is reaching the breaking point,” Mierswa said.
RapidMiner, which has its U.S. base in Cambridge, Mass., raised $5 million in Series A funding in November 2013 from four investors. Last June, it acquired Radoop, which makes it easier to use RapidMiner’s analytics services across enterprise Hadoop environments.
So far, the company has more than 250,000 active years, including teams at companies including Lufthansa, PayPal, PepsiCo, Sanofi and Volkswagen.
In mid-September, RapidMiner appointed a new president and chief operating officer to round out its leadership team. Michele Chambers was previously with MemSQL, Revolution Analytics and Netezza (acquired by IBM for $1.7 billion). Mierswa, a data scientist by training, started developing RapidMiner while he doing research with the Artificial Intelligence Division of the University of Dortmund in Germany.
There’s a limited free version of RapidMiner’s service, but if you want to go beyond CSV and Excel data sources, you’ll need to pay. Pricing starts around $1,000 for an annual personal subscription.