The art of big data jiu jitsu
The biggest challenge when working with big data is often establishing what question is the best one to ask, Andy Jones, who heads the risk engineering practice at MS Amlin, said yesterday during a workshop at the annual Airmic conference in Harrogate.
“The challenge is identifying what you want from the data, otherwise you will get lost.” - David Dexter
In a session called ‘Data Jiu Jitsu’, a workshop hosted by MS Amlin, Jones explained how the firm is working in a number of ways to bring some of the data and analytics expertise it now has in-house to its clients.
He said that for MS Amlin to remain relevant to its clients, this would become increasingly important going forward.
“We are still learning, but the use of data analytics and manipulation will become the norm as time goes on,” he said.
“It will become more and more a central part of our lives. There will be things such as more sensors in the workplace, and data will increasingly become a new language of the workplace. We will need to understand how best to understand and interpret that.”
Jones said that MS Amlin was focused on three main potential uses of data. The first is as a predictive tool, whereby it might be used to avoid future losses and flag up warnings that things could go wrong.
The second is as a validation process, where data will be used to look backwards and better understand why something happened. Finally, it will increasingly be used to establish operational efficiencies, helping companies become more efficient and get answers faster.
“The key challenge is establishing what are the right questions to ask,” Jones said. “That can be the single most difficult thing. A step change is coming for all of us—in fact it is here now.”
David Dexter, casualty risk engineering manager covering casualty and motor, MS Amlin, used the analogy that big data is like teenage sex.
“Everyone is talking about it and thinks everyone else is doing it, but no-one really understands it or knows how to do it,” he said.
He explained how very large datasets can be analysed to reveal patterns, trends and associations in many instances, and offered some insights on big data.
Dexter noted that there will be 30 billion connected devices by 2020, and that 5 quintillion bytes of data are produced every day globally, but that less than 1 percent of that is used in a meaningful way—according to Harvard Business School.
He agreed with Jones. “The challenge is identifying what you want from the data, otherwise you will get lost and drown. There is often so much information that you have to break it down into smaller decision-making processes,” he said.
Dexter also emphasised the importance of being able to present data in interesting ways that others can understand.
