Science not size

Use and abuse
Kieran Towey is a management scientist in Accenture Ireland who reckons that “Data science along with Big Data and Analytics are all terms that are broadly used and abused. I see data science in origin coming from two directions — traditional maths and stats and people with great skills in manipulating and modelling data for targeted outcomes. That strand is now combined with computing technology to bring us to data science, which is not a taught discipline so it is still largely experience-based and combinative. People like myself who have been doing it for years know what works and what doesn’t.

“So in practice data science is growing from a small base of people applying their knowledge in places like, for example, Google or other broad-based organisations and sectors like banking. Big Data and analytics are widely misunderstood and it is imagined that there is some kind of silver technology bullet available. There is not. What we do have are advanced and sophisticated tools — for humans to use. The applications and objectives are real life things like better decisions, reducing fraud, improving healthcare. Personally, I’m inclined to the view that our ICT hardware has made enormous forward strides while the software has in truth lagged behind significantly.”

Applied analytics, Towey says, will always be multi-disciplinary with broadly three strands — technology, data science and specific domain knowledge. “Without that last element, I simply don’t know what is novel or a useful insight. But with my tech colleagues I hope we can make visible the patterns in the data so that the domain expert can say ‘All of that is expected but this one here is interesting so let’s pursue it further. The insights we acquire are always gateways but it is always the domain experts, the clients in our case, who control the process and decide what to explore and in what priority. It is a central involvement, not an analytics exploration where we go off for months and eventually come back with something that may well mean nothing in particular to the potential users.”

 

advertisement

 

Business lead
A similar view from a somewhat different perspective comes from another Irish consultant Alistair Herriott, head of information management and vice president of IASA Ireland, a branch of the international association for IT architects. “I have been seeing the leadership in what we can call data science coming from the business side, people who have an involvement in or responsibility for applied business analytics. Many of them are in fact looking to polish up their ICT skills with college and other courses. They certainly see career advancement as a potential result and experienced business people with both domain and technology knowledge as the new leaders in applied analytics. In general I think the technology people are inclined to look at the go-faster data crunching possibilities rather than focussing on the actual problem or opportunity in practical business terms.”

Big Data and social networking and all of the other masses of data generated about people and businesses are making popular media headlines and Herriott believes strongly that there are serious issues beyond the hype.

“Today we have vast quantities data available and a diverse set of data analytics tools and techniques with the ability to provide real actionable insights that can benefit society. Yet as I see it, we are held back, justifiably, by the lack of an international shared vision for ethics and governance around data. Doors could open properly for the scientists if they could work within a clear set of ethics that is compatible with our culture and values. If they had unrestricted access to the data, with the supporting governances in place, they could provide benefits in many areas including healthcare, civic planning and education.

“This would require clear governance with trusted regulation to protect individuals from a privacy and human rights perspective and from a commercial and data ownership point of view. At this early stage this means that society needs to be included in the discussion and setting of in order to create the understanding of what will be done with the data. This has to go beyond the current scope of data protection that focusses on silos of data and individual organisations. To be really as useful as its potential, Big Data needs to review data across a broad and rich spectrum of data sources. But in truth society in general is now losing out because of a lack of clear ethical data governance in regard to how all of the potentially valuable data can be gathered and used.”

Faustian pact on privacy
This theme was taken up by Dr Barry Devlin, one of the foremost authors and experts in business insight with perhaps the longest pedigree in data management. His new book is “Business unIntelligence—Insight and Innovation Beyond Analytics and Big Data”. A founder of data warehousing and author of the first book on data mining when in IBM Ireland in the Eighties, he is now CEO of 9sight Consulting, based in South Africa. “Today there are quite simply wonderful things we can find and analyse from today’s growing wealth of data. But there are also huge risks in terms of our rights to personal privacy and anonymity. The sad truth is that we have lost most of our privacy already in a Faustian pact to share, often unwittingly, details of our daily lives. Facebook ‘likes’ and Google ‘plus’ may be a conscious trade-off between a loss of privacy and a perceived increase in social capital or useful contextual information. Yes, our smart phones report our location minute by minute — but we may be offered a reward coupon for a nearby retailer at any moment.”

Devlin is also slightly sceptical about the novelty of much of what is taken to be modern data science. “Power users in data mining, business intelligence and analytics have for many years been doing what is largely what is now grandiosely termed ‘data science’ — even if they were often using Excel as the immediate tool! What is hugely different is that in the past the data they were using came from business systems designed for specific uses. Everyone knew what the data meant. Now the data is coming from outside the organisation and global sources. Some of it is human sourced information — Facebook, Twitter, blogs and audio and video — that is by comparison inherently unreliable, even pure fiction! Another growing source is machine-generated data, the Internet of things, with the smart phone as the current primary species of thing. Now smart phones generate lots of information, like location, but there are many assumptions underpinning any use of that information — such as that the device is with the owner.”

“We are combining data collected for one purpose and utilising it for other purposes and so in constant danger of drawing inferences that are really not supported. Whether you call them power users or data scientists, that human ability to distinguish dross from useful information is vital as well as an understanding of the relative truth or validity of different data,” says Devlin.

The key areas of priority research include:
• The Semantic Web
• Sensors and the Sensor Web
• Social network analysis
• Decision Support and Optimisation
• Connected Health

The Insight mission statement is clear: “We live in a data rich world, where almost everything we do results in a data record being created and stored somewhere. Our vision is to create a healthier, safer, more productive world by empowering a data-driven society to enable better decisions by individuals, communities, business and governments.”

As a national research platform it is intended that Insight will become a single entity and brand with the critical mass of research competence that will serve as an international beacon for the science and application of Big Data Analytics. Insight combines the skills of leading researchers with cutting-edge technologies from diverse fields of research. It works closely with industry partners to develop next-generation data acquisition and analytics solutions for many key application areas.