Mediaplex_tag

Willis and Towers Watson have merged. Visit willistowerswatson.com

Rise of the data scientist

Colin Gibson | 20th April 2015

TECHNOLOGY

Many companies are applying data science as a fundamental part of their sales strategy - techniques employed online to try to influence our buying behaviour. The role of the data scientist in most organisations involves constant innovation and experimentation, refining models and looking for new needles in new combinations of data haystacks.

Voters Like You Also Voted For.....

With the UK general election campaign now in full swing, you can be sure that each of the main parties has teams of data scientists, the alchemists of the 21st century, trying to turn torrents and lakes of data into political gold.

Pollsters and pundits have been trying to predict the outcome of elections for years. This has always involved the use of data (opinion polls) and statistical analysis. Only recently has it attracted the label “data science” as so-called Big Data has made it possible to perform ever more complex calculations on increasingly massive sets of data. It hit the headlines with the US Presidential election in 2012 when the American statistician and writer Nate Silver not only predicted an Obama win when most other pundits were saying the outcome was too close to call, but correctly predicted the result in every one of the 50 states and Washington D.C.


Influencing the outcome

Being able to predict the outcome of an election with confidence is useful, particularly if you are on either side of a betting relationship. But campaign managers need more than that. What do they do if their confident prediction is telling them they are going to lose? While Nate Silver was stealing the limelight, Obama’s campaign was applying data science at the most granular level, predicting which voters were “truly persuadable” using an analytical technique known as “uplift modelling”. It enabled the Democrats to focus their efforts and resources on the voters most likely to be swayed (strictly speaking, those more likely to be positively than negatively swayed) by a leaflet drop, phone call or canvasser on the doorstep.

In the UK, the Labour and Conservative parties have both enlisted heavy-hitters from the successful Obama campaigns of 2008 and 2012, David Axelrod and Jim Messina respectively. It is safe to assume that they will be looking to reuse any tricks and techniques that translate to the UK environment, and that these will include the wizardry of the data scientists. They are not likely to broadcast their activities ahead of Election Day, but once the dust has settled I am sure stories will emerge of the role that Big Data, data science and predictive analytics played in at least one of the UK Parties’ campaigns.

The volume of data collected and processed during a single Formula One race is greater than that in the US Library of Congress.

The rise of the data scientist

Before data science hit the mainstream, there was a saying that “A data scientist is a data analyst who lives in California” – reflecting the role of the tech giants of Silicon Valley in birth of the discipline. If that discipline ever really was confined to California, it has now spread far beyond its borders. Demand for data scientists is booming. A recent Harvard Business Review article labeled the role as "Sexiest Job of the 21st Century?" McKinsey have predicted that, “By 2018, the United States alone could face a shortage of 140,000 to 190,000 people with deep analytical skills”.

Using models to assess how different decisions and actions could affect future outcomes is not new. One of my first jobs, way back in the last century, involved the use of computer simulations – war games – to assess the performance of air defence system designs. Quantitative analysts – “quants” – in banks have for years been modelling the performance of exotic and not-so-exotic financial products.

What has changed is the array of computing tools and the universe of data that can now be exploited. In the right hands, these can enable more detailed models to be developed and more certainty in predictions to be achieved. Trends and correlations can be spotted in data sets that could not previously have been handled or, in some cases, did not even exist. The volume of data collected and processed during a single Formula One race is greater than that in the US Library of Congress. Most of the racing now happens in cyber space.


In the mainstream

Of course the demand for data scientists is not being driven by election campaigns, a topical but somewhat niche area. Many companies are applying data science as a fundamental part of their sales strategy - we are all used to the “Customers like you also bought….” techniques employed online to try to influence our buying behaviour. The role of the data scientist in most organisations is not cyclical as it is in politics. It involves constant innovation and experimentation, refining models and looking for new needles in new combinations of data haystacks.

At Willis, modelling and analytics are key to helping our clients understand and manage their risks. Some of our tools will be showcased at the RIMS conference in New Orleans later this month. As companies recruit their own data scientists will this reduce their need for our expertise and advice? I don’t believe so. They will be looking to exploit Big Data and data science in product innovation, getting ahead of the competition and increasing sales. Risk management is important, but providing advice on risk management solutions is not a core competency of most organisations

Companies like yours work with risk advisors like us.



About the author

Colin Gibson

Colin Gibson is Willis’ Global Architecture Director, responsible for all aspects of the company’s enterprise architecture.

  • Copyright © 2019 Willis Towers Watson