Hi ,
When you're first learning data science, building machine learning models from scratch is a great way to develop an understanding of how they work "under the hood".
Which is one
reason so many machine learning courses take a "build first" approach.
Yet, the problem with this approach is that data scientists come to feel it is somehow "wrong" to consider anything other than building from scratch - to the detriment of their future employers.
To quote data engineer and friend-of-list Andrew Jones, who recently wrote about this topic in the context of engineers:
"Building everything ourselves is costly, takes a lot of time, and requires ongoing maintenance."
This answer is to "give yourself permission not to build".
Here's the thing...
Your job as a data scientist isn’t building models. It’s using data
to solve business problems for your employer.
And in most business situations, the best solution is also the one you can deploy the fastest.
If a pre-built model exists that will solve the problem, that will almost always beat anything you could build
yourself.
Talk again soon,
Dr Genevieve Hayes.
p.s. I recently had the opportunity to discuss the "build vs buy" decision with North Labs CEO Collin
Graves, in the context of building your organisation's data science capabilities.
You can listen to our conversation at the link below: Episode 42: Should You Outsource Your Data Team?
