Using scheduling algorithms to reduce infection exposure in the workplace
Here at Rotageek we use algorithms to create schedules for our customers as part of our Autoscheduler product. We already take into account lots of different considerations about what makes a good workplace rota, from making sure that employees get enough rest between shifts, to ensuring that key skill sets are available when needed, and generating schedules that seem fair if you were to compare with a similar colleague.
Given the recent outbreak of COVID-19, we thought we’d see if we could help in a way that utilises our experience in this area. While many businesses have temporarily moved towards a work from home model, which can be great for tech companies where there is minimal impact on the business, for some of our customers, perhaps working in the retail, hospitality, or healthcare sectors, remote working isn’t a feasible solution.
For these businesses, we’ve just released some functionality that tries to reduce and contain the exposure risk inherent in their schedules by updating our algorithms to take this into account. The idea is similar to one that many large offices have recently adopted by splitting their workforce in half, to try and contain any outbreak within that segment, and preserve the other half of the company to continue functioning as normal. While this method may be somewhat effective, it’s a very coarse tool, and we feel there’s a more nuanced way we can approach it using our Autoscheduler.
We’re trying two different approaches to tackle these issues:
The first approach is easy to build with our already existing tooling, it offers an advantage over the current state of affairs quickly, and allows a manager to segregate the workforce into different cohorts, which we then try to make sure only work with each other. This approach allows employers to decide, in advance, which colleagues will work together and guarantee that there is no overlap.
For the second approach, we created a brand new scheduling rule to look at generated schedules and penalise schedules that have large degrees of colleague overlap. This may result in patterns that would be hard to create manually, but preserve more of the other attributes of a good schedule that we know colleagues and managers care strongly about.
This way, within the other rules that have already been set, the rule will try to create a schedule which minimises the number of times you would switch from working with one group (where someone may have exposed the team to a virus) to working with another group (which has not been exposed, and you could be the infection vector to, even if asymptomatic).
Scheduling has always had to take into account many different factors that impact where and when employees should work, and what they should do. I don’t think anyone predicted that we would need to take this specific one into account, but we want to do what we can using the tools we have available to help our customers reduce the risks to employees and their families, as well as managing the impact this has on their business to make sure they still have the right distribution of skilled colleagues to function effectively.
If you have any questions, or would like to use this in your business, reach out to our friendly 😊, real-person support team here, who are on hand to help 7 days a week, just on the other end of that support chat bubble!
Stay safe! 🚀
