In the last years, I felt a lot of admiration and maybe a little bit of envy towards my colleagues and me just because we are the Data Scientists. Yes, it is a hot word, the media is talking all about AI, machine learning and predictive intelligence. Who would not want to know what their future will look like, right?
Data scientists are so hot now because they can predict things. But can they always?
While talking with my colleagues from different industries, we all agree on the same things:
Don’t be intimidated by your data scientist
Most of our colleagues are intimidated by what stuff data scientist talk about.
Reinforcement learning, neural networks, math, statistics, Natural language processing (NLP), clustering, supervised and unsupervised learning, alfa, gamma, cohen’s kappa and tons of other terms.
I’ve seen many times words thrown randomly by my peers to make them sound fancy, and I wasn’t proud of that always.
Don’t be intimidated by the terms your data scientist is using it. Be curious, ask your data scientist: What’s a neural network? Or why do you use five clusters?
There is no shame in asking. Only pretending that you know what it means.
Data scientist doesn’t know all answers all the time
If the data scientist says it, it must be the truth.
This is another thing that happens in most companies and most teams.
If we work with numbers and we are the predictive wizards, doesn’t mean that we are always right. I’ve seen many times; Data Scientist copies and paste old projects for a new problem they try to solve. The reason can be: very tight deadlines, high pressure or just only because of that is what they all know.
In the end, they will come up with numbers that they will swear that this is the ultimate truth. But not always that will be the case.
You as a domain expert should always ask your data scientist why they think that that approach is the best and why the numbers they present are the ultimate truth.
The data scientist can be good with numbers and predictive, but he will never be as good as a domain expert. By asking the right questions, you both can come to an even better solution than it was before.
Data scientist can fake it too sometimes
Unfortunately, many of my colleagues’ data scientist I have talked to experienced it at least one. There will always be a person that wants to fake it until he/she makes it.
They all know the terms and the techniques but fall to realize it in practice.
If you don’t want to fall for their trap, ask them more questions.
In the end, all solutions that your data scientist gives you may look fancy and cutting edge, but never forget to ask them: Why do you think this will work?
What’s your experience and thoughts?