The Data Science Unicorn is a fantastic beast. Much like the regular unicorn, the data science unicorn is expected to ingest through the horn and output a multi-hued spectrum of insights. But unlike the regular unicorn which no one expects to meet in the real world, many real-world businesses expect data scientists to be unicorns – a magnificent creature who is data engineer, analytical programmer, business analyst, systems engineer, financial expert and then some. Add in a few more technical buzzwords like IoT, Python, Hadoop, Tableau, AWS and the job ad looks like it came out from the rear end of a unicorn. It just does not work that way and leads to frustration all around. How to avoid this? That’s the discussion in this post.
Why do data scientists quit?
An inspiring article by Jonny Brooks-Bartlett points out the main reasons why data scientists are job hopping. Really, there’s only one reason – they’re unhappy with their job.
- Data scientists quit because they are expected to wear many hats. Anything that is quantitative or technical in regards to data gets foisted on the data scientist.
- Data scientists lack the infrastructure to be truly productive or are expected to build said infrastructure. There’s a difference between a data engineer and a data scientist.
- Data scientists are expected to do ETL (Extract, Transform, Load), and set up your big data infrastructure, which leaves little to no time for actual DATA SCIENCE.
- Data scientist are usually an isolated unit within the company, thus their input is very limited and is constantly being questioned about their value.
When are data scientists happy?
Work isn’t work if it’s something one enjoys doing. There are mundane tasks in every position, but hopefully, the mundane elements are less than 20% of the role. A team member is only motivated if there is
- clarity in job function,
- alignment with job expectation, and
- empowerment in the working environment.
People. Processes. Technology. An alignment across all three makes for a healthy work environment. So you need a good leadership team, and the business needs to be truly driven off data. If your business rates 1 or 2 on the analytical maturity framework, [Download free analytical maturity calculator here] then you are not ready to hire a data scientist.
Tools of the data science trade – an environment that empowers
- Data Mart: You need an environment for collecting, cleaning, storing and managing data. This means a (semi-)automated feed from all data sources, processes to clean and upload data into the centralized repository, and processes for managing data integrity, security, and privacy. The data scientist will need direct access to this data mart.
- Access protocol: The access to the data mart can have checks and balances but should not be constrained by IT. In my experience, if the data requests get funneled through an IT specialist, it hinders the analytical process. Data scientists, in particular, if they are handling an open-ended project, need to explore different analytical paths. Choking off data access just hamstrings the team.
- Analytical tools: You would not paint a house with a toothbrush and a data scientist is not going to get anything meaningful done on a spreadsheet running on a stock computer from Best Buy with 4GB RAM. I cannot be prescriptive since needs and tastes differ. Big data needs specialized tools. More on that in the next section.
An environment for data scientists
Unsure where to start, look over the Polytab proposition. It’s the turnkey customer data platform with a built-in analytical workbench and web application tool.
Where to start? Download the product flyer
So if you’re a data scientist and need to influence your CTO, or you’re a CTO looking to jump-start your data science function, or you’re the CXO looking to drive your business with big data analytics, you need to consider the Polytab data management platform with the analytical workbench.
Use the web form below to get the product flyer. Let’s talk soon.