Issue 17

A day in the life: statisticians

We talk to four statisticians working in different jobs across industry and academia

post

22 May 2023

Statistically speaking, most Chalkdust readers hope to find a job involving lots of maths when they grow up. (Statistically speaking, so do most Chalkdust editors.) There are so many different ways to do maths, it can be overwhelming to think about the different options. Luckily, Chalkdust is here to help. This is the second in our Day in the Life series, where we ask mathematicians to tell us about a typical day in their line of work. This time, we’re digging into the data-day lives of four people who work with statistics:

  • Helen Parsons, a medical statistician at Warwick’s Clinical Trials Unit.
  • Rory McLaurin, a higher statistical officer at the Office for National Statistics.
  • Vilda Markeviciute, a machine learning specialist at Expedia.
  • Evangelos Sariyanidi, a principal data scientist at the Children’s Hospital of Philadelphia.

All four of them use statistics and data to make the world better, from fixing shoulder injuries to helping people to book their holidays. Their jobs involve maths, coding, data analysis, and lots and lots of coffee. If that sounds like your thing, read on—you might be inspired.

Helen Parsons

I’m Helen and I’m a medical statistician. I’m an associate professor and I work at the Warwick Clinical Trials Unit as a trial statistician. When I was thinking about what to write here, it really is the stock statistical answer: “it depends”. My days vary immensely, not only because of the work, but also because I work flexibly across the university campus, a local hospital and remotely at home.

Helen tries to remember her password

Typically, I get into the office bright and early to deal with the first task of the day: email and planning what I need to get done today. On today’s hypothetical agenda, I start off with meetings. As a trial statistician, my meetings are many and varied: firstly, there are the trial meetings to make sure that the clinical trials are running well—the safety of participants is our number one priority—but I also keep an eye on anything which could affect data integrity and the trial analysis. Secondly, there are collaboration meetings. These include talking with clinicians to design and plan new studies, and with other methodologists to talk though ideas and problems. Finally, there are ‘catch up’ meetings. This is mostly me checking in with students and the junior staff that I manage to make sure that everything is OK. However, if I’m at the hospital offices, my first meeting of the day is always a fun one: sitting in the cafe and having a weekly catch up with all of the team. I find these informal meetings really valuable, as I can catch up with the trial management team’s queries, ask the clinicians to explain details about data I don’t quite understand, do the reverse with an explanation about stats, and occasionally, discuss the latest binge-worthy boxset.

In-poster syndrome

Soon enough, it’s time for lunch. Sometimes that means a sandwich at my desk, sometimes it’s going to (or giving!) a research seminar, but I also meet up with friends around the medical school and catch up. With hybrid working it takes a bit more planning than it used to, but it’s always worth the effort.

After lunch, today’s job is to analyse the trial data and see if the intervention has an effect. This is definitely my favourite part: it’s the culmination of years of hard work and putting theory into practice is always a fun challenge. Plus, being the trial statistician means that you are the first person to know what the trial result is, which is awesome… but also means that you need to have a good poker face until the results are made public! Then, before I know it, it’s time for one last check of the email to make sure I know where I’m working tomorrow, then it’s time to log off and head back home.

You can read a report Helen helped write here.

Rory McLaurin

I’m Rory, a higher statistical officer at the Office for National Statistics. I work in the business statistics transformation division of the ONS, where we try to improve the quality of the surveys we send to businesses and the data and statistics we produce from them.

The start of my day typically involves a ‘coffee and coding’ catchup with whoever is around. Sometimes someone will bring a thorny problem along to try and get some insight from others, but more often than not this is just a nice way to start the day and get a sense of how everyone’s doing. After that, I try to get my admin tasks out the way: replying to emails, scheduling meetings, and planning out the rest of the day.

By the time that’s done, I’m awake enough to get on with some proper data analysis. I work on some projects individually, but I most enjoy the collaborative sessions where I get together with a colleague to share ideas. A lot of our projects involve measuring or predicting the impact of changes to our surveys or data processes, so we’re usually working with data from a long time period. That means a lot of data cleaning, and then some conceptual work trying to turn a question like “was this change good?” into a measurable statistic we can produce. When we’re collaborating remotely, tools like Git for sharing code, Teams for screen sharing, and even Paint for drawing quick diagrams, make a huge difference.

Rory on the beach

In the afternoon, we often have a meeting with another team to share updates and get advice. The teams that actively look after our database, design new surveys, or process new data are all fantastic at helping us figure out weird results or suggesting useful areas of analysis we might have missed. Sometimes we invite subject experts from across the organisation to these meetings to help us understand the wider context behind our work.

Finally, towards the end of the day, I tend to get some focused individual work done. Depending on where I am in a project, that could be some background reading, writing new code, or writing a report on my results. The ONS publishes a lot of reports on all the work we do, which you can find online. One of the best things about my job is that it’s pretty flexible—so if I don’t have any meetings or urgent publications, I can finish early and go surfing.

You can read a report Rory helped write here.

Vilda Markeviciute

My name is Vilda and I work as a machine learning scientist at Expedia Group in London. I’m a part of the reinforcement learning team, helping different parts of the business build adaptive experiences optimised for every traveller!

My journey to machine learning started with a few side projects which sparked my curiosity for what an ML/data science job in industry would be like. That’s how I came across a summer internship at Expedia, where I discovered interesting ML problems, innovative technical solutions, and a supportive learning environment. I enjoyed it so much that I decided to join the company full time after completing my PhD. (I still remember thinking I’d never become a data scientist while working through my first ever Python tutorial, but look how things change…)

As Expedia has a hybrid work policy, I go into the office 2–3 times a week. Having the option to work from home helps me balance the in-person collaboration (nothing beats a good whiteboard session) with focus time at home. A typical day in the office looks something like this:

8am: I catch a train to King’s Cross. Good time to respond to emails or read some papers if I’m feeling awake, day-dream otherwise!

9am: I reach the office, go straight to the coffee machine, get some cereal and catch-up with my colleagues.

Mornings are usually focus time for me. Some days, I might be reading papers looking for ways to improve our current multi-armed bandit algorithms or to solve new problems. Other days, I might be coding up those new algorithms to be used in our in-house Python simulation package. I recently learned Scala, which is used in our production environment, and I’m glad our team has a bunch of engineers who are always more than happy to help if I get stuck. Sometimes, I deep-dive into data from the latest use case or simulations using Spark.

12:30pm: it’s time for lunch! If it’s a sunny day, we might stretch our legs on a walk to the nearby Exmouth market. We then go to the roof terrace in the office where we eat and chat about everything from our weekend plans to the latest ML trends (ChatGPT has been a popular topic recently).

Vilda, not on the beach

After lunch the meetings begin! As we work with people from all over the US and Europe, the afternoon is the time our workdays overlap. Some days, I would be catching up with the teams using our reinforcement learning platform to help them design a new use case or to update them on the progress of an ongoing one. No two are the same, so it’s a great way to learn about different parts of the business and meet people working on a wide range of products.

2:30pm: we have our team’s daily meeting. We give updates on our work in progress and discuss any urgent issues.

3pm: time for our weekly journal club! This time, our lead ML scientist is giving a tutorial on neural Thompson sampling. After almost two hours of discussions and reading, my brain is melting, but I leave the room inspired and with a long list of references I want to revisit.

5:30pm: I catch up with some emails and messages, check up on the progress of the simulations and prepare for tomorrow.

It’s time for the pub, or for the train home!

Evangelos Sariyanidi

I’m Evangelos Sariyanidi, a principal data scientist at the Center for Autism Research (CAR) of the Children’s Hospital of Philadelphia. Our goal at CAR is to develop new technologies for diagnosing autism faster, earlier, and with higher reliability. We aim to bring the machine learning revolution into the methods we use to diagnose and treat autism—we want to replace the pen-and-paper tests and human observation with algorithms that automatically measure and quantify behaviour directly from videos of people. The goal is to find methods that are less time-consuming, and also more precise.

At CAR, we can use existing computer vision (CV) algorithms for a lot of things—but it’s my job to fill in the gaps. Specifically, I’m researching new methods that are needed for behavioural research, such as taking videos of people’s facial expressions and head movements and finding ways to quantify them. As well as conducting research, my responsibilities include writing software, seeking out opportunities for funding, and communicating with other research groups to set up new collaborations.

Headphones optional

Today, I’m writing a research paper with my colleagues. These are usually engineering studies about new CV methods, or behavioural studies based on applying CV to clinical research.

We’ve spent some time on a literature review, reading the most recent papers about automated behaviour analysis and how it can be applied to autism and other disorders. We’ve also completed a plethora of experiments, which have created lots of data for me to curate—that means programming. Depending on the specific paper, I can find myself coding in Python, Matlab, C++, or even—like this morning—Cuda. Once the analysis is done, I’ll meet up with some other engineers to discuss the results and see if we can find ways to improve on them. Later on, I’ll chat to some clinicians and behavioural scientists to decide on the next analyses we should conduct, or the next CV methods to investigate. We also work to turn our CV methods into more finished software, so that other engineers and behavioural scientists can use it too.

After lunch, I work on a research proposal. We need funds to keep the centre running, and we are constantly on the lookout for new funding opportunities. Writing a proposal requires plenty of internal discussions, doing some preliminary analysis to justify that the funding is needed, and then some highly collaborative writing sessions before we can send it off. I trade emails about our results to other research groups, so I can learn more about the research they’re doing, and hopefully form new collaborations.

Finally, I log off and head home for a well earned rest.

Do you have a mathematical day job?

Get in touch—we’d love to feature you in the next issue of Chalkdust!

More from Chalkdust