What type of data scientist are you?

A look into the most common data scientist personas

Data science is a relatively new field that has incorporated profiles and personalities from several disciplines. But what are the different types of data scientists and which jobs were these people doing before the advent of data science?

Most people have seen some form of this trilogy chart, that I have re-adapted below.

Successful data science is indeed a mixture of three core components: statistics, business acumen, and programming skills, but few people are strong in all these areas.

So, which persona are you? Or, if you are trying to hire a data scientist, which one do you need?

Three background paths lead to three different personas:

• The recovered management consultant

• The researcher escaped from academia

• The developer turned data scientist

The recovered management consultant

This category spans the junior business analyst and the ex McKinsey consultant. They have in common a passion for Excel and their ability to show off v-lookups and fancy formulas even to plan their house move.

They are also the ones who have more passion for the business problem: they come business first, data after.

They had to learn Python or R by necessity, not because they enjoyed programming. And they still try to avoid coding as much as they can and their code is generally as re-usable as a single-use napkin.

They have good intuitions for the basics of statistics but they had to learn concepts like p-value or t-test the hard way.

What they are good at: data science projects that support decision making, business-oriented processes, one-off projects.

The researcher escaped from academia

They often have a PhD and come from a research background. They studied hardcore math and statistics and they could speak for hours about the philosophical differences between the Bayesian and frequentist approaches.

They are normally ok at coding, as long as they don’t have to push themselves too much into the boundaries of data engineers. A test-driven programming approach might be a stretch for them.

But they are probably good at lower level programs such as C++, which could come handy for applications at large scale or deep learning.

What they tend to lack is business thinking. Developing a product is probably the end goal for them because they recognize that as the equivalent of publishing a paper in academia.

What they are good at: complex machine learning projects at the front edge of innovation. They can push boundaries, read lots of research papers to pick and implement the best ideas. A deep-tech company would probably need a handful of those profiles.

The developer turned data scientist

Since data science requires a lot of coding, these are probably going to be your best friend.

You can trust them to build good reusable code, don’t have to explain to them the concept of testing, and they will probably be able to automate the pipeline more than you hoped for.

They are probably fine to use machine learning tools out of the box, but if they need to venture into deeper statistical thinking, it could become a minefield.

Some of them might be decently good at understanding the business side of things, particularly if they were previously involved in gathering requirements and handling business relationships. But don’t expect them to be too proactive with novel business ideas.

What they really excel in is a scenario that requires technical challenges, such as a big data project that needs to work at scale or a complex data pipeline. At the same time, they are less comfortable in a project which requires more advanced modeling and where prediction accuracy needs to be a competitive advantage. They are also best suited for projects where they have to focus on building products rather than on supporting decisions.

Conclusions

If you are a data scientist or are thinking to become one, try to figure out which persona you fit in and which are your gaps, but also don’t be shy to recognize your strengths.

As a data scientist, you will need a mixture of all these skills, but expecting to be a pure all-rounder is a myth: there is always a side you prefer!

And if you need to hire a data scientist, try to stereotype which category the candidate fits more into: hiring someone to build a high-scale recommender engine will certainly require a different profile than someone to support your CEO in sales forecasting.