R number modelling hackathon

Hackathon

As part of the fortnight event series “Actuarial Innovation in the Covid-19 Era” we invited you to take part in our R number modelling hackathon. Here are the results.

Hackathon - The question

“Perform your own critical evaluation of the R number, demonstrating potential pitfalls and explain your conclusions to policymakers (a non-actuarial audience).”

More detail on the activity can be found on the ICAT Hackathon specification.

Hackathon - The winner

After careful consideration, the judging panel decided that the team led by Krishna Kumar Shrestha was the winner. 

"This team produced an analytically sound and clear analysis of the precarious situation in Nepal. Their output was careful in its description of the analysis and the data limitations. Its final communicate clearly articulates the crisis in Nepal and its relationship to neighbouring India. It represents strong recommendations for the national disaster management. It thus well covers the key items we were looking for – appropriate use of data; sound analytical methods; appropriate evaluation of the analytical results and their limitations, and finally strong communication of the results and their meaning. The judges commend the team for their bold work."

Hackathon overview

The exact problem statement was delivered to contestants at the start of the hackathon which took place Thursday 29 April – Tuesday 4 May. The general area of the problem was an extension of the R analysis to accommodate emerging issues, in a similar way that the ARG October bulletin developed additional insights that could have further shaped public understanding and policy. 

Background

Early on in the pandemic, the world got to understand the concept of the R value of a pandemic development. This is the number of people that each infected person infects. An R of less than 1 means that over time the pandemic goes away, a number above 1 means that more and more people are infected over time. This has become a common number used by the scientists and government to describe the current situation and therefore the reason for lockdown and social distancing action. Indeed COVID-19 can be compared to other diseases and pandemics using this R number. In the UK the ONS publishes data weekly on an estimate of the R number and it is widely reported.

But actually the R number is too simplistic – should you look at the R number for a region or country or continent; how do you calculate the R number in real time; what differences does an R number mask that would be more or less important for policy makers?

The Actuaries Response Group (ARG) has been at the forefront of examining and explaining the issues as they arise. But they also highlight the issues that are still to come, or worse, are missing some crucial pieces. In their bulletin of 18 October 2020, they analysed the issue of superspreaders. In this analysis they questioned the R number metric and described circumstances in which it is more or less useless as a diagnostic metric or policy shaping metric.

It is this type of issue that the hackathon addressed. What additional insight is needed on key items that are discussed in public, and used by government to make decisions? 

The activity

Working in small teams, teams performed the following activities:

1. Analyse relevant data: for the R number evaluation this would be UK ONS COVID-19 R development reports and their underlying SARS-COV2 prevalence studies, but you could use alternative data, or supplement this data if you choose to.
2. Performance analysis: build models and develop analytical tools to understand the drivers of current developments and project these into the future
3. Draw conclusions: provide results and conclusions to prove or disprove a current hypothesis on what is happening and what may happen next
4. Communicate the results: report to a knowledgeable non-actuarial audience

Hackathon rules

In participating in this challenge, you agree to the following:

1. The hackathon is open to anyone aged 16 or over, excluding employees of the IFoA and their families.
2. Entries are limited to one submission per team.
3. At least one member of the team is an IFoA or affiliate member.
4. The IFoA reserves all rights to disqualify you if your conduct is contrary to the spirit or intention of the challenge.
5. The panel will choose the winning team from all, eligible entries received.  The decision of the panel is final and no further correspondence will be entered into.
6. By submitting an entry you confirm that the entry is your own original creation.  You agree by your entry submission that the IFoA can use your idea, findings or methodology as it considers appropriate for the use of the whole profession and you further agree that the IFoA may reproduce the entry for publicity in relation to this and/or other promotions or initiatives.
7. You will not use any third party material, data or information which you do not have authority to freely use without limitation and will not breach the intellectual property rights of any third party in your participation.
8. No entries will be returned.

Missed out or considering a similar activity?

If you have any questions regarding the hackathon, would like to provide further insight to the question or are interested in doing similar activities please contact Katy Stephenson