10.03.2020

IFoA Certificate in Data Science – Colin Thores interview

Wikipedia[1] defines data science as:

“ … an inter-disciplinary field that uses scientific methods, processes, algorithms and systems to extract knowledge and insights from structured and unstructured data. Data science is related to data mining and big data.”

It feels like “data science” has always been there. Many of my actuarial peers have been excited about “data science” and “big data” for a number of years and the recent BBC panorama documentary, “Amazon, what they know about us”[2], suggests that the tech giants embraced data science a long time ago.

APR has worked on a number of data science projects, with clients both inside and outside of our traditional industry sectors. Whilst as a firm we recognised the opportunities for actuaries to make a valuable contribution in data science projects some years ago, I personally have been a little slower warming up. However, as I started to research the topic, I began to appreciate both the personal development and commercial opportunities which data science provides for actuaries and actuarial firms.

For this article, I interviewed Colin Thores, Education Actuary at the IFoA, about the new Certificate in Data Science course (the ‘Certificate’) which he has developed over the last 18 months and which launches in April.

Chris: Colin, why are the IFoA creating a ‘Certificate in Data Science’?

Colin: Data science, in all its manifestations, is rarely out of the public eye these days. Advances in computer science, statistics and data science means that data can now be collected, stored and analysed much more quickly than in the past. Increased digitalisation offers new data sources which can be combined with traditional types of data. This gives the ability to process and draw useful insights from data in a way that was not possible even a few years ago. These factors are transforming most major industries.

In particular, data science is increasing in importance in all the main areas that actuaries work. The growth of big data and data science techniques potentially presents many opportunities for actuaries. The contribution of the actuarial profession could be uniquely valuable to the emerging field of data science given our public interest mission, our strong professional standards, our grasp of the statistical underpinnings of data science, and our ability to interpret and construct value from data.

There are also potential threats to the future relevance of the profession if actuaries are not able to participate in data science related work.

There has been considerable discussion within the IFoA and with actuarial employers about what data science education for actuaries should look like. It is clear that members have different needs with regards to data science. Not everyone in the profession needs to be a data science expert or an ace programmer but members do need to have an understanding of the core data science techniques and language, and an understanding of the potential issues for consideration in applying them in an actuarial context.

Whilst there are an increasing number of actuaries who have an understanding of data science, the general level of data science knowledge across the IFoA membership is currently quite low. John Taylor the current president of the IFoA has made increasing understanding of data science across the actuarial profession a key aim of his presidency. The IFoA’s Certificate in Data Science is therefore intended to raise the general level of knowledge about data science across the IFoA membership.

Chris: I share your general view that actuaries are well placed to add value in the data science field. Taking a balanced view, are there challenges or difficulties actuaries could face in the field?

Colin: Yes, I think there are a range of challenges for actuaries in data science. There is no doubt that there will continue to be advances in computer power and the amount of digital data available in the future. This will impact all businesses, but it is particularly likely to impact data heavy businesses of the kind that most actuaries work in. For actuaries to be credible in data science related work we need to continue to build our understanding of the language, concepts and techniques in data science. I don’t think as a profession we can or need to be experts in every aspect of data science. But given the complexity of some data science work, I do think we need to build stronger relationships with other professionals and be prepared to work with them.

I think there will be further challenges around ensuring that as a profession we act in a way that retains public trust with regards to data science work. Given the importance of ethics and professionalism in our profession I think we are well placed to do this.

Chris: Could you outline the key elements of the Certificate Colin?

Colin: The Certificate is fully online so it can be accessed by all IFoA members regardless of where they are based. The Certificate is module-based and will cover the following topic areas:

Core data science content.
Introduction to data science and techniques.
Application of data science techniques to solve real business problems.
Data visualisation and communication.
Machine learning and artificial intelligence techniques.
Hands-on experience of some of the tools used widely in data science (e.g. Python, Tableau).
Application of data science applied in traditional and non-traditional areas of actuarial practice.
Actuarial related case studies.
Content related to the practical, regulatory, professionalism and ethical issues of using data science techniques in an actuarial context.

The Certificate is stand-alone and will sit outside the IFoA’s current Fellowship qualification. The IFoA felt that it was important that every IFoA member who wants to access the course can do so regardless of where they are in their career and so the target audience for the Certificate is all IFoA members. In terms of requirements to take the Certificate we have assumed a basic knowledge of statistics but there is no minimum requirement for programming knowledge.

The Certificate will be awarded jointly by IFoA and the Southampton Data Science Academy (SDSA). SDSA is a partnership between the Web Science Institute (at the University of Southampton) and global education provider Cambridge Education Group. The course is assessed. Each attendee must complete three assignments and obtain an overall average of 65% to attain the Certificate. More details can be found here: https://www.actuaries.org.uk/learn-and-develop/lifelong-learning/certificate-data-science

Chris: What sort of time commitment is required?

Colin: The course will be run on a cohort basis. Each cohort will run for 10 weeks – 8 weeks supported learning plus two additional weeks to allow completion of the final assignment. The first cohort will launch in April 2020, and there will be 2 further cohorts over 2020. Based on feedback we will decide how many cohorts will be run each year thereafter.

We expect the course will require roughly 80 hours of study. This is obviously a material commitment. We are trying to balance the needs of the individual with putting enough information in the materials to make it worthwhile, and without making it excessive, which would put off many members. There are a lot of practical data science related exercises within the Certificate, using tools such as Python and Tableau. Individuals can choose to do these or not – they are not assessed. We do however think members will gain a lot from doing these exercises. To put this in perspective, we would expect one of the IFoA’s Specialist Principles (SP) exam to take roughly 200 hours of study. We expect the Certificate to be materially less challenging than an SP exam.

Chris: I appreciate the balance you are trying to strike Colin. For members who can’t commit 80 hours, are there resources available from the IFoA which provide shorter CPD opportunities?

Colin: Absolutely! We are looking to provide a range of different data science related CPD opportunities for members including:

Online member resources – e.g. online data science related professionalism videos.
The Guide on the Ethical use of Data Science in collaboration with the RSS.
Data science related conferences.
Involvement in different streams of the Data Science Member Interest Group.

Given the interest across the profession in data science related material the IFoA will definitely be producing more data science content in the future.

Chris: How is the course landing with members?

Colin: The level of interest from members has been fantastic. Members have been able to record interest in the Certificate for a number of months now on the IFoA website, and we have received over 3,800 notes of interest from members. The IFoA ran a webinar outlining the details of the course on 26 February and 180 people attended the webinar. Members could apply for the Certificate from 24 February 2020 and the first cohort (which was limited to 100 people) was fully subscribed on day one. Once we have run the first cohort, we will look at the feedback provided by the attendees and develop the course materials and course as appropriate.

I’d like to also mention the support we’ve had from a wide range of members and employers in developing and reviewing the content for the Certificate.

Chris: Is data science something that is better suited to newer entrants to the profession rather than the more experienced members?

Colin:That’s a difficult question to answer. I really don’t think you can say data science is just for the new entrants and not for more experienced members of the profession. I think it is possible members could be involved in data science related work in different ways depending on the requirements of their job and the stage they are at in the career. For example, it is possible newer entrants to the profession could end up doing more of the coding and analytic work earlier in their careers. More experienced members, that have built up deep subject matter expertise, could find themselves acting as subject matter experts within a multidisciplinary data science related team, with the analytic work being carried out by others. That is a pretty sweeping generalisation though!

I think data science will increase in importance for the profession as a whole, and it will therefore benefit all members to develop their understanding of data science techniques, regardless of where they are in their careers. Depending on the requirements of an individual’s job, I can see some people will need to focus on those particular aspects that are relevant for them at that point in time. We are going to keep a close eye on who is sitting the Certificate to try and understand why they are taking it.

Chris: Colin, can you tell me a bit about the data science ethical landscape and how the profession is contributing to that debate?

Colin: The ethical use of data is a cornerstone of the actuarial profession and there is already a range of guidance to support members around this. The increasing use of date science does raise potentially challenging new ethical issues which the actuarial profession needs to be aware of. The IFoA has developed additional guidance for its members with respect to data science related work. The guidance is set out in A Guide for Ethical Data Science (the ‘Guide’). The Guide is intended to complement the IFoA’s existing ethical and professional guidance and is intended to be used as a practical framework for actuaries to consider the ethical issues within a data science related project.

This Guide is available on the IFoA website here: https://www.actuaries.org.uk/upholding-standards/standards-and-guidance/professional-standards-directory. The Guide was developed following a collaboration between the IFoA and the UK Royal Statistical Society (RSS), and launched in Autumn 2019. The Guide draws on both organisations’ commitment to ensuring high professional standards and upholding the public interest. I understand the IFoA will look to develop further guidance around data science in the future.

Chris: You and your colleagues at the IFoA have certainly been busy over the last 18 months, what’s next for the IFoA in terms of data science education?

Colin: We think that data science will have an even greater impact on actuarial education and professional development in the future. With regards to the Certificate, we intend to review the material regularly. Once we have feedback from the first couple of cohorts, we will ensure this feedback is used to update the Certificate materials accordingly. We intend to develop more technical content if that is what people need.

With regards to the IFoA’s Fellowship examinations we have already added the use of R and some aspects of data science (e.g. Machine Learning) for the recent amendments to the Actuarial Statistics exam in the 2019 curriculum changes. Over time, we expect more data science related content to be added into various subject in the Fellowship exams. However, we need to be careful that we keep the amount of materials at an appropriate level for each exam to ensure students are not overloaded.

Lastly, we expect the IFoA to develop more CPD events, webinars etc to support Certificate holders and the membership in general.