Sergey Nivens - Fotolia

Data scientist vs. data analyst: What's the difference?

Data scientists and data analysts have a lot of crossover in their roles, but they're certainly not the same. Here's a look at some key differences in the positions.

There's a lot of hype around data scientists today, but the reality is that many companies are still in great need of data analysts, too. Data analysts play a key role in helping business users keep their eye on the ball and solve day-to-day problems. In many ways, they can complement the work of data scientists, but they're also important -- even critical -- when companies don't have a data science program.

It's useful to consider the data scientist vs. data analyst differences so enterprises can build the right team and individuals can hone the most appropriate skills.

"Data analysts place an emphasis on inspecting and analyzing data [and] creating reports, while data scientists focus on experiments, research and machine learning," said Ji Li, data science director at Clara Analytics, which provides an AI platform for the commercial insurance industry.

Same principles, different questions

The data analyst and the data scientist use many of the same principles, often work on similar sets of data, address similar questions and face similar hurdles in their work. A core data scientist vs. data analyst difference is that analysts are usually given a set of questions they need to answer, while data scientists are usually expected to ask their own questions, said Kirill Eremenko, founder and director of SuperDataScience, an AI educational service.

Analysts excel at looking at data to find previously unseen patterns using descriptive and diagnostic analytics. Conversely, a data scientist attempts to identify patterns in data sets and then uses those patterns to predict how the data is likely to behave in the future using predictive and prescriptive analytics.

As the jobs are so closely knit, a data analyst is in a great position to develop further into a data scientist.
Kirill EremenkoFounder and director, SuperDataScience

"As the jobs are so closely knit, a data analyst is in a great position to develop further into a data scientist," Eremenko said. However, this will require data analysts to change their approach, he said. They would have to learn skills around forming their own hypotheses based on the data they have available and then either prove or disprove those theories.

Eremenko started out as a data analyst at Deloitte Analytics in Australia. At Deloitte, he predominantly used data to answer questions such as "What happened?" and "Why did it happen?" He subsequently got a job as the first data scientist at Sunsuper, a pension management company, where he had to test out various algorithms for predictive and prescriptive analytics.

His role at Deloitte would sometimes task him with addressing more open-ended questions, such as "What will happen?" and "How can we make it happen?" However, these questions were either clearly defined by the managers and directors or were handled directly by them. Thereby, the managers and directors were doing most of the critical thinking that data scientists typically are responsible for.

Analysts understand the business

Data analysts tend to be closer to the business users and tend to be experts of the data at hand, said Rosaria Silipo, principal data scientist at Knime, a data science and analytics platform. They know the business case, the data collection process and the data domain in and out. "They might not be mathematicians, but they can offer great insights on how to acquire and handle the data and how to interpret the results," she said.

There is often a fair amount of overlap between data analysts and data scientists. Both process data with in-depth domain knowledge and math expertise. Over time, Silipo finds that experienced business analysts can sometimes grow their knowledge in statistics and machine learning to improve their value. At the other end of the spectrum, data scientists and engineers can learn more about the data collection process and business cases, especially after a few years in the field.

Head chefs and line cooks

Cheryl To, data scientist at ThinkData Works, a data wrangling tools provider, said one useful metaphor to understand the data scientist vs. data analyst difference is that data scientists are the head chefs, while data analysts are the line cooks. Head chefs are capable of what line cooks can do, but they should be coming up with the overall menu and theme of the restaurant and meal. Line cooks are specialized in doing the necessary prep work and gathering the ingredients needed for those meals.

She said, "Most often, data analysts are tasked with a specific problem where they will leverage data to derive a meaningful solution." This complements the work of data scientists, who have more freedom to explore and generate their own questions based on their analysis.

For example, at AI Foundry, a mortgage automation tool provider, data scientists play a key role in developing the company's deep reinforcement learning and cognitive business automation platform, said Peter Piela, director of development at AI Foundry. His team of data analysts perform a variety of tasks related to collecting, organizing and cleaning the data to assess quality and trends. This team includes business analyst specialists who assist in testing activities and research in order to understand lending document automation issues. They also work with data curation specialists who apply an attention to detail to prepare the model training documents.

"It is the business domain knowledge that the data analysts offer that is invaluable to the data science team," Piela said.

Cultivate new skills

Data analysts can differentiate themselves by honing a variety of technical and soft skills. Vivek Ravisankar, CEO and co-founder of HackerRank, a developer hiring service, recommended analysts focus on improving their understanding of statistics and data wrangling, particularly using tools like Python and R. It's also important to master visualization and dashboards in tools like Tableau, Looker and Excel to provide insights and effectively communicate with key stakeholders.

He also recommended that they be aware of new technologies and markets that can impact the kinds of data that companies find valuable. For example, IoT data was not as valuable a few years ago as it is now.

Both data scientists and data analysts will often present data findings to internal stakeholders. Consequently, they must both have the ability to relate their work to diverse audiences, said Dr. Angel Durr, CEO and founder of DataReady, a data literacy program. "Good storytelling and organizational skills are an essential aspect of both careers," she explained.

Additionally, both data analysts and data scientists must be comfortable with a high degree of ambiguity. They need to learn how to manage and maintain data processes effectively and document processes for the purpose of constant process improvement and development.

Durr recommended analysts cultivate some level of CRM expertise, since most organizations use these systems in combination with other sources to see the overall picture. "Understanding data and understanding the specific needs of your domain area of expertise will make you invaluable to any organization," she said.

Dig Deeper on Business intelligence management

Data Management
Content Management