Data scientists are in high demand among businesses, ranging from startups to Fortune 500 companies, with the growth in demand for data scientists expected to rise by 19% from now until 2026 – 12% higher than the norm.1 Data scientists require a unique combination of technical acumen and insightful analysis: part mathematician, part computer scientist and part trendspotter, they bridge the gap between business and the world of IT.2
Top roles and responsibilities of a data scientist
A data scientist’s main role is to organise and analyse large portions of data through custom-designed analytical software, in order to provide stakeholders with findings that they can use to make informed business decisions.3 A data scientist typically has various roles and functions. However, there are some common responsibilities:4
- Gather large amounts of unstructured and structured data, and transform it into a more understandable format
- Utilise various programming languages, such as SAS, R, and Python, in order to extract value and insights from the data
- Identify trends and patterns in data that may improve a business’s profitability
- Find answers to business-related problems by using data-driven techniques
- Communicate and collaborate with IT and the business as the point of contact
- Stay up to date with analytical techniques, such as machine learning, deep learning, and text analytics
The key functions and tasks of a data scientist
There are typically four key functions of a data scientist within an organisation, with two aspects to differentiate job functions: ‘for’ vs ‘as’. If the data scientist is there to support a team that is building something, then ‘for’ is applied. If they’re there to build something themselves, then ‘as’ is applied.
A data scientist may be required to create something that is customer-facing; this is called a ‘product’.5 Alternatively, the data scientist may be required to build a backend system that is vital to the business running smoothly; this is called ‘operations’.
1. Data science for products
(Inference scientist).6 The data scientist provides insights to improve products and company strategies. This could be for high-level strategy, or a more hands-on tactical analysis regarding the performance of a specific product. The following skills are required to perform this function:
- Exploratory analysis.7 Using scripts and programming language SQL, a data scientist explores and summarises data sets to answer key questions about the product, such as: What unique behaviour is important to track product health, and can we identify the factors that are linked with this behaviour?
- Business intelligence.8 Data scientists are not always limited to providing insight specific to the data. They can also be asked to advise on how the business should react to the data, and their role shifts into more of a business intelligence function, which will then require more collaboration with operations
2. Data science as a product
(Applied scientist).9 In this function, the data scientist uses data to improve business products that are customer-facing, as well as using machine learning to make data products that support customer-facing endeavours. Some key skills to carry out these tasks include:
- Machine learning.10 A working knowledge of how to incorporate machine learning’s ability to apply algorithms to large sets of data from alternative data sources – such as text, images, and video – and to learn from it, and forecast trends, will give data scientists more insight through predictive modelling
- Prototyping.11 Data scientists should be able to design and prototype machine-learning software products before going ‘live’. By creating a minimum viable product, they can assess if a company should allocate resources into building out an entire system
- Software engineering.12 When it comes to implementing something that will affect an enterprise, they need a reliable and repeatable process. An approach that is well-structured, while remaining agile, is needed to streamline the graduation of an asset from development, through the test phases and staging process, and on to production. It typically involves people from many cross-functional roles. Understanding system programming languages allows the data scientist to build and release data products that are adaptable and can be rolled out in various environments. By using software engineering best practice, a data scientist will ensure that the software programming used for data products is robust, reliable, and maintainable
3. Data science as part of operations
(Machine learning engineer).13 In this function, the data scientist is typically a member of an engineering team, whose objective is to build non-customer facing data products that are necessary to run the business. This function may require the following skills:
- Online learning. The nature of real-time system requirements means that standard batch learning is not viable, and online methods for updating models should be utilised
- DevOps. The data scientist who is utilised in DevOps will need to know how to create and maintain data products upon which business functions operate
- Communication.14 The data scientist will need strong communication skills to explain the strategy and motivation for a project plan with business owners
4. Data science for operations
(Systems scientist).15 In this role, data scientists are required to use root-cause analysis to determine the breakdown of system performance. Model building will assist to gain greater insight into how various factors, both internal and external, impact on systems. Here are the required skills for this function:
- System understanding. A holistic understanding of the systems and infrastructure used to build products will facilitate insight into how different factors influence operational metrics
- Forecasting.16 In order for a data scientist to identify anomalies in data, knowing how to establish a standard data baseline and then forecast expected data behaviour is a key requirement
- Alerting.17 In this role, a data scientist may be required to ascertain when it’s necessary to communicate with other teams about unusual system behaviour
A data scientist is the common link between systems and business, and is fundamental to any business that wants to achieve and sustain a competitive advantage. Combining data, computational science, and technology with consumer-oriented business knowledge, a data scientist’s functions and tasks revolve around driving and producing high-value insights for decision makers within the business. Playing a leading role in project management for projects that require large volumes and variety of data to be processed, the data scientist creates solutions that result in enhanced overall business performance.
- 1 (Nd). ‘Computer and information research scientists’. Retrieved from BLS. Accessed 20 May 2019.
- 2 (Nd). ‘What is a data scientist?’. Retrieved from SAS. Accessed 22 April 2019.
- 3 White, S. (May, 2018). ‘What is a data scientist? A key data analytics role and a lucrative career’. Retrieved from CIO.
- 4 (Nd). ‘What is a data scientist?’. Retrieved from SAS. Accessed 22 April 2019.
- 5 (Nd). ‘Data scientist’. Retrieved from Cleverism. Accessed 7 May 2019.
- 6 Weber, B. (Feb, 2018). ‘Functions of data science’. Retrieved from Towards Data Science.
- 7 (Nd). ‘Data scientist’. Retrieved from Cleverism. Accessed 7 May 2019.
- 8 (Nd). ‘Data scientist’. Retrieved from Cleverism. Accessed 7 May 2019.
- 9 Weber, B. (Feb, 2018). ‘Functions of data science’. Retrieved from Towards Data Science.
- 10 Talagal, N. (Apr, 2019). ‘Hiring a data scientist: the good, the bad, and the ugly’. Retrieved from Forbes.
- 11 Talagal, N. (Apr, 2019). ‘Hiring a data scientist: the good, the bad, and the ugly’. Retrieved from Forbes.
- 12 (Nd). ‘Data scientist’. Retrieved from Cleverism. Accessed 7 May 2019.
- 13 Weber, B. (Feb, 2018). ‘Functions of data science’. Retrieved from Towards Data Science.
- 14 Talagal, N. (Apr, 2019). ‘Hiring a data scientist: the good, the bad, and the ugly’. Retrieved from Forbes.
- 15 Weber, B. (Feb, 2018). ‘Functions of data science’. Retrieved from Towards Data Science.
- 16 (Nd). ‘Data scientist’. Retrieved from Cleverism. Accessed 7 May 2019.
- 17 Talagal, N. (Apr, 2019). ‘Hiring a data scientist: the good, the bad, and the ugly’. Retrieved from Forbes.