The impact of data science on enterprise operations is everywhere to be seen. Affordable, high-performance computer power brings data science applications within reach of the commercial mainstream. Digital disciplines once regarded as rarefied specialisms are bursting onto digital transformation agendas across all vertical sectors.
Actuaries, whose professional skills are predicated on the analysis of data to assess and manipulate outcomes, occupy a unique position. Their job roles in many ways anticipate the outcome-shaping capabilities data science provides and puts them in pole position to turn disruptive change to their ultimate advantage.
Similarities between actuaries and data scientists mean that there’s been growing consideration of how – and where – the two skillsets intersect and the effect that will have on actuaries’ career journeys.
The IFoA recognises this intersection. It has already acknowledged its importance to members in a series of events, institutional initiatives and published guidance. And it has identified a need to provide a point of focus for members and actuarial affiliates who realise that data science will – to some extent – shape our profession as we move into the 2020s.
This content is designed to meet members’ information requirement with respect to data science and its resonance for actuaries and actuarial practice.
Its aim is to bring together the IFoA’s multifarious work and activities in the data science disciplines and highlight the Institute’s understanding of the opportunities data science presents for actuaries to extend their professional status and societal responsibilities.
For such a transformative concept, data science is often not well defined or understood. This does not seem to inhibit its influence – IFoA President John Taylor has described it as necessarily a ‘dynamic and evolving term’.
However, a working definition for data science (rather than of data science) will prove helpful as a start point.
‘Data science’ describes a broad, multidisciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and insights from structured and unstructured data. It employs techniques drawn from many fields within the context of mathematics, statistics, computer science, and information science.
The ‘data sciences’ include, or associate with, more focused disciplines: Big Data, Data Analytics, Data Analysis, Data Mining, Artificial Intelligence, Machine Learning, Robotics, Data Visualisation, Predictive Modelling, and Deep Learning.
Data science skills combine domain expertise, programming, software tools and knowledge of mathematics and statistics, to extract desired insights from data – insights that can be translated into tangible and quantifiable business value, such as market intelligence, risk assessment, and executive decision-support.
“Data science is very empowering for actuaries: it gives them a platform to move into wider fields.”
John Taylor, President, The Institute and Faculty of Actuaries
In February 2018, the IFoA launched a virtual conference on the topic of data science. It was open to members globally to share knowledge and discuss developments and techniques within the discipline.
The conference demonstrated how actuaries can enhance their current knowledge, or move more fully into the world of data science. It offered a platform for members and students to learn more about the subject in sessions curated by leading experts.
Topics discussed included Big Data analytics, Artificial Intelligence, Machine Learning, predictive modelling, data visualisation, neural networks, and coding.
President Taylor said: “What makes the contribution of the profession uniquely valuable to the emerging field of data science is our public interest mission, our grasp of the statistical underpinnings of data science, and our ability to interpret and construct value from data.”
President Taylor believes that the scope of the IFoA’s engagement with data science “demonstrates the way in which the actuarial profession is already taking the lead in addressing the issues that will shape our global society, while deploying our skills and knowledge to safeguard the public interest.”
These issues include ethical questions around how data science changes the way an individual’s risk profile is assessed, through to ensuring that ‘data science’ – in each of its applications* – is well used and understood wherever it’s practiced.
Comparisons between actuarial science and data science often use the preposition ‘versus’, although the disciplines are not in necessarily in opposition to, or in contention with, each other.
Indeed, they have much in common, and practitioner skills are increasingly transferable between the two fields.
Usually annexed to the enterprise IT function, data science uses scientific methods, processes, algorithms, and other software tools, to derive insights and knowledge from structured and unstructured data sets.
It brings together established disciplines like statistics, data mining, and data analytics, with emergent tech like Artificial Intelligence (AI) and Machine Learning. Its capability to process a broad range of data sets, and not be constrained to specific types or formats, is generally acknowledged as a key strength.
Actuarial science is a more formalised discipline that categorises interrelated fields, such as computer science, economics, finance, mathematics, probability theory, and statistics.
It’s applied to defined mathematical and statistical methods to assess risk, primarily in the BFSI (banking, financial services, insurance) sectors. Actuarial science is, however, also applied in other industries and professions.
Experts in data science and actuarial science use many of the same techniques when analysing data to make informed forecasts about risk probabilities. These include data visualisation, pattern recognition, and statistics. Three examples of where the two disciplines align are in their approach to data, methodology, and software development.
The primary commonality between data science and actuarial science is their respective appetites for data. The more data they have, the better their analyses turn out.
The IFoA already provides data science collateral for members that ranges from:
“Immersed in business context, actuaries are well-placed to utilise the insights data scientists can generate.”
John Taylor, President, The Institute and Faculty of Actuaries
It’s sometimes asserted that actuarial science specialises in structured data, while data science is most adept working with unstructured data. However, the insurance sector has long recognised the importance of unstructured data.
DataSpace cites the importance of vehicle telematics for data scientists and actuaries alike. For example, telematics provides data that enables data science optimise performance of autonomous cars, while insurance providers have identified an opportunity to use telematics data to adjust premiums based on driver behaviour (‘usage-based insurance’).
Actuarial science’s problem-solving methodology is driven by the standardisation of its analyses, while data science focuses on getting the prediction correct through a variety of methods. Mastery of actuarial methods is largely based on qualified skills and certified learning. Data science practitioners may work without mandatory qualifications or certifications, although this anomaly is changing.
Actuarial science operates on pre-developed software development platforms, and generally uses standard commercial tools and applications. Data science differs in that it routinely develops new algorithms for specific use-cases. Data scientists are, perforce, likely more ‘programming savvy’ than the actuarial scientist.
