Data is the oil of a digital age. How many times have we heard that? Despite becoming a bit of a caricature, Data is a massive industry that is only going in one direction: up. And despite the pandemic putting the breaks on many industries growth plans, data, on the other hand, has only had its importance highlighted with smart, remote working becoming mainstream.
As the demand for data grows, so too does the demand for professionals that can analyze and process voluminous amounts of data to derive valuable business insights from it.
But what makes an effective Data Scientist? We brought together a few Data Scientists to discuss their role, what they believe makes an effective Data Scientist and what they think the future holds for the role.
This is the first instalment in a series of articles that focuses on the careers and experiences of data professionals across the Nordic region.
Data Science is a versatile field that applies processes, concepts, tools, technologies, and theories to empower the extraction and analysis of valuable data to derive meaningful information. In other words, data science is widely practiced to help businesses make data-driven decisions. Data Scientists are the one who practices data science. Some of the techniques involved in data science include data extraction, data mining, data retrieval, and data analysis. A data scientist can add value to any organization in many ways.
Data Scientists are like a trusted advisor to a firm’s management and play a vital role in any organization as they educate them on the value of the companies’ data to enhance the decisions taken by the upper management.
With this wide range of remits that change depending on the industry and company a Data Scientist might work within, it’s hard to give definitive answers on what makes an effective data scientist. The best we can do is bring a few professionals working in that space together to discuss what they believe makes an effective Data Scientist.
Brian Ye is a Data Scientist at VISMA, a B2B software company. Based in Stockholm, Brian’s daily tasks involve analysis of smart services, using data collection platforms and working with public cloud technologies.
Having worked at VISMA for over three and a half years, Brian works on VISMA’s data science tasks, problems and goals from forecasting in the workforce management domain to setting up data collection solutions for services.
Domain Knowledge is Key
The term “Domain Knowledge” has been in play even before data science became popular. In software engineering, it means the knowledge about the environment in which the target (i.e. software agent) operates.
We can use the same definition in data science to say — “Domain knowledge is the knowledge about the environment in which the data is processed to reveal secrets of the data”. In other words, the knowledge of the field that the data belongs to is known as Domain Knowledge.
For Brian, one of the most important aspects that a Data Scientist needs to be effective is good business and domain understanding.
“Data Science can be a costly discipline. So, it’s important to get knowledgeable about your business domain so you can be more efficient in getting to the core of the problem and delivering the right solution quickly.”
You may have studied data science and machine learning and used some machine learning algorithms like regression, classification to predict on some test data. But the true power of an algorithm and data can be harnessed only when we have some form of domain knowledge.
It’s understandable that many data scientists (among others) object that agile’s philosophy is only for software development. After all, software development teams have been the overwhelming driving force to agile adoption. And even the Agile Manifesto itself is titled that “Manifesto for Agile Software Development”.
However, agile’s adoption has increasingly grown beyond just software development, and ironically, many of agile’s roots don’t come from IT (Harvard Business Review, 2016). Non-software implementations of agile include: National Public Radio in programming creation, John Deere for new machinery, Team Wikispeed for electric cars, Saab for fighter jets, C.H. Robinson for human resources management, and Silicon Valley Data Science for train timetable predictions.
For Brian, the next five years will be telling for Data Science as it looks to diversify the range of tools used to deliver data science as quickly as possible.
” I think going forward, there will be a high focus on how to enable tools and how we can adopt developer practices and DevOps practices.“
Arunabh Singh is the Director of Data Science at small FinTech company Willapay. A US focused invoicing and payments app for freelancers, Arunabh works in a team of around 15 people and is solely responsible for the data science and analytics but is looking to grow his team over the next few months.
Arunabh’s personal advice for those looking to improve their Data Science game is to be obsessed with solving problems that your business has and then find the tools to solve them – not the other way around.
“Don’t start with the fancy tech. Understand the problems first.”
Secondly, Arunabh suggests focusing on how you talk about the work you do.
“Be as a good of a storyteller as you can be because no one is ever just convinced by the facts.”
Changing a stakeholder’s mind requires a lot more than just data. It requires a certain level of imagination to take data and turn it into insights and then take that insight and turn it into a compelling story.
“Data should be used as an argument to a good story that you’re already telling.”
According to Keynote Speaker Bernard Marr, “You can invest in data technologies and collect all the data you can possibly imagine, but it’s worthless if it’s not analyzed or communicated to decision-makers so that action can be taken from the insights”.
A data scientist must have intellectual curiosity and a drive to find and answer questions that the data presents, but also answer questions that were never asked. Data science is about discovering underlying truths. Successful scientists will never settle for “just enough”, but stay on the hunt for answers.
“The best data scientists, in my opinion, are curious. They look beneath the hood of problems, and are self-driven to find out more information that is missed at a cursory glance. So the best data science insights are clever, but counter intuitive. If it’s too intuitive, it’s then it’s probably you’re missing something,.There should be an aha moment later that, okay, I tried to slice and dice this differently. And this is what it gave me. And now that I reevaluated, it makes sense, it should add to the existing pool of knowledge, rather than just reinforcing biases.”
Currently, there are no formal routes to becoming a Data Scientist. You need to have good quantitative skills an aptitude for problem solving and the ability to communicate in a business development context. Beyond that, however, there are no existing structures or paths to take.
For Arunabh, the future will bring more specialist branches of data science in a similar fashion to software development with different languages and specialisms.
“I would like to see more specialisation within the data science that people do. So like, some people who might come from marketing will be focused on marketing data science. And the tools become more and more easy for them, it will be easier for like juniors without advanced technical skills to pick up and specialise in that domain.”
Data Scientist, as a role, has been around for almost 12 years now. That’s enough time for those who entered the space at the start to be heading towards very senior positions. Arunabh is curious to see how data science will be represented at the executive level and if that transition is already happening.
“We have seen that happen with other domains, but with data scientists that are only just getting to that point. So I’m curious to see if like, any of the big fortune 100 companies will have a CEO who started out and worked his way up his or her way up as a data scientist.”
Nicolas Innocenti is a Data Scientist at a newly started digital identity solution, Svipe. With Svipe from the very beginning, Nicolas’ remit as a Data Scientist extends to backend software development and AI work. With an educational background in Physics and Informatics at KTH Royal Institute of Technology, we were keen to see what Nicolas believes makes an effective Data Scientist.
It doesn’t matter how good you are at the algorithmic side of data science – if you don’t understand your data set (where it came from, what it means and how it’s generated), you won’t get very far.
“Because if you don’t understand what the data means you don’t understand the features of it, then you can be as good as you want at the algorithmic part, it won’t help.”
Nicolas, who works in the digital identity space, is well versed when it comes to understanding and interrogating his data set. On top of that, he also believes it’s very important to understand the value of the question as well as the answer.
“For example, we need to detect if people have glasses on the photo to ask them to remove the glasses. Yes, you could do it with deep learning, gathering 1000s of pictures of people with glasses, but is it worth the effort for what it brings to the app? So I think there’s always that sort of metrics which some new people struggle with. We should always interrogate solutions to see if they make business sense.”
Nicolas, who has been interviewing Data Scientists for years now, believes that Data Science – and the people involved in that space – is moving the emphasis away from softer skills in favour of algorithmic and technical skills.
“I’m a bit worried about Data Scientists’ profiles. When we were recruiting data scientists in 2016 it was like looking for Unicorns. Now I can see a clear trend that has data science focus on optimising an algorithm.”
Believing the technical side of the role being algorithmic, Nicolas is worried that other aspects of the role will go neglected.
“When you focus so much on the technical side, it’s much more difficult to get a good understanding of what you’re doing. Focused on using an algorithm to get the metrics as high as you can isn’t going to solve the business problem. So if you don’t have an understanding of the relation between what you’re doing as a business problem, I don’t see how this can work.
And I’ve seen those new titles appearing that so AI engineer, BI engineer, Deep learning and ML Engineer . I mean, what are all those roles? This should be one person, really.”
Magnus Gidlund is the Lead Data Scientist at Nordic Entertainment Group. With a focus on analytics products, Magnus’s role is to support other analysts and data scientists on his team, further develop the collaborative culture within the team and align the data and analytics team with the strategic goals of the wider business.
One of Magnus’ more prominent tasks is the development and constant improvement of an analytical framework designed to help the team interrogate the questions and requests they receive so five different people aren’t answering the same question in five different ways.
“We’re in a big organisation, and a lot of people are working with analytics. So I think it’s very important to have these frameworks, and then it’s, it’s a part of my role is to kind of help building that.”
Data Scientists come from all walks of life. Some will focus on the engineering side, others on the software side. Some, like Magnus, will be focused on the statistical side of data science.
“My profile is trying to translate business problems into something we can quantify so it’s important I have a solid understanding of statistics.”
But that isn’t the golden ticket for Data Science as many of their projects rely on explaining and translating this statistical information into something a Leiman can use.
“Working as a freelance musician, as I did for eight years, helped me build up that mentality of being able to explain things. But you don’t have to work eight years as a musician to follow my footsteps. The key is having that mindset of teamwork. If you work with someone, you want it to be smooth and kind of easygoing process. And I think that’s, that’s important.”
Mikael Huss is a Senior Data Scientist and Co-founder at Codon Consulting. With a team of four, including three founders and an employee, Mikael is currently looking at expanding the team as their project portfolio builds.
Currently working on a few different Natural Language Processing projects including a project to flag potential dangerous messaging sent through an app, Mikael and the team at Codon Consulting are working non-stop.
We were glad to be able to get the chance to ask Mikael about Data Science as a career, what he think makes an effective Data Scientist and what he thinks the future holds for the job.
Having an ‘always learning, always curious’ mindset is a key skill that Mikael values highly. As far as dealbreakers go, Mikael believes that communication is top of the list when it comes to skills a Data Scientist needs.
“It’s very important to be able to explain to your clients, whether they be internal or external clients, what you are doing and not get lost in technical details. So it’s very nice to be able to give a presentation that can capture the imagination of those who are listening.”
The role that emerged several years ago looks very different to the role we have today with many companies honing in more clearly on what they want from a Data Scientist.
“It was supposed to be someone who both knew, statistics, programming and had domain knowledge. But now, the roles have become more specialized: data scientists, machine learning engineers, data engineers, and in some cases, maybe even data translators and roles like this. I think the data scientist role is going more towards being able to link to stakeholders and consulting with them on data.”
With his current project around NLP, Mikael is using a wide range of different tools. This makes a change for someone who is used to inventing and implementing algorithms from scratch to solve a problem. For Mikael, this is the future.
“It becomes more about selecting the appropriate tool, knowing what you can do with all of these tools, and being proactive and suggesting to your stakeholders what they should do. I think there will still be data scientists, but they will be at a somewhat higher abstraction level where you write less and less code, and it will mostly be calling different API’s and different packages and knowing what they are good for.”
