Data Scientist Where are the imposters?
My post will be a little bit provocative… yet, it is MY opinion that there is a common misconception about who and what data scientists are.
Whenever I read through a post, either on LinkedIn or other media, written by a self-proclaimed Data Scientist, I instantly become weary. Indeed, this gives the false impression that the data science community is larger than it really is. As a result, hiring real data scientists with strong skills and acumen has become a true challenge.
So, my question is down to earth: Do we have numerous imposters or simply a misunderstanding?
Thru its expertise, Business & Decision Switzerland has identified three types of profiles. It is a pragmatic approach and not scientific one (no complex algorithms needed).
1. Real Data Scientist. They are few and have strong skills and knowledge in programing, mathematics, statistics, machine learning…
2. Citizen Data Scientist (a concept introduced by Gartner). The idea is to leverage internal skills to bring the basic data science expertise in-house for advanced analytics while minimizing the burden on organizational resources. Citizen DS enjoy using graphic user interface (GUI) developed by software providers that embed advanced algorithms.
3. Data Engineer. Here are the imposters that artificially enlarge the data scientist community… The data engineers spend most of their time preparing data and the rest modeling. They can therefore not be considered real data scientists, but due to the complexity of today's data architectures (through variety, velocity and volume, tricky subtilties around architecture, in memory computations, distributed data and so on), the data engineer is crucial to any data science project.
The following graph shows the key differences in time split between these three profiles:
The key messages that we therefore want to provide is: Increase the efficiency of your data science project with two complementary profiles.
This combination will allow the data scientist(s) to focus on extracting value from your data with innovative modeling approaches while capitalizing on data engineer's expertise.
At Business & Decision Geneva, our data science project highlights these two complimentary profiles.
Data & Digital Architect | Consultant
1 年Thank you for sharing, Yann! ??
Principal Data Scientist, Healthcare Artificial Intelligence (AI) CoE at Fidelity Investments
6 年Mostly true, but there can be overlap between DS and DE. A model is only as good as the input data, good features often have a much larger impact than the latest and greatest modeling methods. I wouldn’t expect someone without modeling experience to necessarily build the best features for me and would prefer to take some responsibility for this. Web apps some BI tools etc may also fall into either role in some cases. Generally the terms are poorly defined in implementation and will continue to change as technology advances and new tools are instantiated.