Key skills for aspiring data scientists: Communication
This blog is the first in our ‘Data science skills’ series, which takes a detailed look at the skills aspiring data scientists need to ace interviews, get exciting projects, and progress in the industry. You can find the other blogs under the ‘Data science career skills’ tag.
If you’re looking to become a commercial data scientist – one that helps organisations actually use machine learning as part of their essential business operations, instead of working on new solutions in the lab – then communication is going to be key. Why? Because, as a data scientist, you won’t be a back-office worker. In a field as complex as data science, your ability to communicate insights and updates could make or break the success of a project.
Why is communication an important skill for data scientists?
Communication skills will be essential in almost all aspects of your data science career, but they’ll come in especially handy for a few key reasons.
Data science is a team sport.
If you’re coming out of a PhD, you’re probably used to a more academic style of working, where each team member works independently or in a small group to tackle their own sub-project. All of this feeds into the success of the main project of course, but it’s a fairly linear process and collaboration is often restricted to very specific stages.
But collaboration is built into the fabric of commercial data science projects. Without it, everything falls apart very quickly. In industry, you’ll be spending a large proportion of your time brainstorming, experimenting, and implementing – and then, in many cases, returning to the drawing board to brainstorm again when challenges arise.
Crucially, this collaboration won’t just involve other data scientists. Teams, unlike most research groups, include people with different backgrounds and skills. To succeed, you’ll need to be able to communicate insights in a way that your non-technical team members understand and can act upon. Your project manager needs to stay on top of timelines and budgets and make sure your solution squares with the requirements. Your team needs to be able to communicate the value of the work to seniors. To do this effectively, you’ll need to be able to clearly articulate updates, insights and potential challenges that arise from the data science work.
Data science insights need to be actionable.
In data science, there’s no use in generating insights for insights’ sake. If a customer (or colleague) is going to pay for a data science solution, then the executive team, frontline staff or other end users must be able to understand and contextualise the outputs of a model. And, if they’re going to actually use model outputs to inform business decisions or processes, they need to be able to understand them fast – without gaining a PhD-level understanding of data science first.
That’s where you come in: it’s your job to explain, not just what a model says, but the implications of the models’ outputs for the business. I worked recently on a fast-paced project that built a text analysis model designed to inform our customer’s response to the COVID-19 pandemic. With the situation evolving every day – every hour, even – we needed to produce usable insights quickly and regularly. The ability to translate technical model outputs into non-technical, accessible information that senior decision makers could process quickly was vital.
Communication is a vital interview skill.
This one is something of a no-brainer: communication is vital for the job, so it’s vital for interviews, too. Whether you’re in-house or part of a specialised data science company, your interviewers will be looking for signs that you’ll be able to mesh well with other team members, work closely with end users and generally make data science accessible.
Remember, you might not always be interviewed by other data science specialists – if you want to give non-technical interviewers a good idea of how your work and why you’d be an asset to their team, you’ll need to be ready to explain past achievements in layman’s terms, too.
Communication helps you get recognition for your work
But why can’t data scientists just leave the customer communications to the project management team or senior management? Managing the needs and expectations of customers or other stakeholders in the company is a major part of their job, after all.
Well, first of all, you’ll still need good communication skills if you want to effectively brief your colleagues on your needs, progress and potential roadblocks. But being able to communicate directly with your customers or other stakeholders in your business also saves time, reduces the risk that detail will be lost in translation and – perhaps most importantly – means you’ll actually get direct recognition for your work. Showing a customer the value you add to a project and knowing that they recognise your valuable input is much more rewarding than being seen as an anonymous, silent cog in a machine. And a bit of recognition from senior clients or stakeholders can do wonders for your career progression, too.
How do you develop communication skills for data science?
Unlike other skills, there aren’t many courses available for developing your communication abilities as a data scientist. Instead, it mostly comes down to practice, feedback and experience.
If you’re still in academia and in the early stages of your journey towards becoming a data scientist, then your best bet is to start discussing your work with non-technical people in your life. Even if your current project or thesis isn’t data science-based, start trying to explain your work to friends, housemates, or parents with no experience in your field. If they understand it well enough to accurately explain it back to you, then you’re on the right track; that means your colleagues and customers will be able to do the same for your data science projects when you’re not in the room. Extra points if you can get your grandparents to understand machine learning.
Or you might want to look into participating in a hackathon in your spare time. In addition to getting to train your technical skills and learn from other data scientists, hackathons are a great opportunity to practice explaining your thinking under pressure and working in teams.
Want to test your communication on a real commercial data science project?
If you’re thinking of transitioning from academia into data science and looking to undertake some in-depth data science training, you might want to consider the Faculty Fellowship: it’s an eight-week programme where you’ll learn both ‘hard’ technical skills and ‘soft’ skills like communication, presentation and customer management. At the end of it, you’ll get to put those skills to test via an industry placement on a real life data science project; that means working hand-in-hand with both technical and non-technical members of your customer’s team to ensure that your models deliver ROI for their organisation.
Head to the Faculty Fellowship page to find out more.