Why Use-Case is king in a Data Science project?
Identification or proposal of a solid use case can massively influence the solution you are delivering to a client. Or for that matter, the in-house analytics team.
I believe in the below equation.
“Use Case:Analytics offering::Tail Wing:Aircraft”
It holds true for a scenario of a vendor offering a solution to a client, or for a company trying to expand out capabilities in data science.
A well defined data science project needs a solid use case. In the current process of data discovery, led by my mentor and assisted by me, we have come across a scenario where our offerings are driving what a potential use case can be. This is quite the opposite of how it is supposed to be. A well defined use case, in my opinion, will help to tailor your products better and identify the right fit to the environment. This is increasingly becoming an issue as more and more companies undertake data science initiatives, in the process of joining the larger bandwagon. Renowned analytics practitioners often reiterate that “Use case is king” and I am kind of experiencing it first hand nowadays.
Building an analytics pipeline with mechanisms to ingest data, perform stuff such as object detection, deploying models on docker containers are all like puppets dancing to the tune of an use case. Without an use case, the bogey will derail. As more and more teams identify avenues for data science ventures, they need to ensure that their use case definition is robust enough. If not, I believe more and more data science and analytics solution providers need to assist the client in channeling the journey and providing a value add by showing what kind of use cases can be pursued.
For in the industry, data science works differently as compared to school. It is much more than writing a script capable of generating text based on RNNs, and submitting it as an assignment. It needs a greater eye, one that am trying to open up currently to inspect what avenues are there to forge a collectively strong partnership.
Cheers,
Kishan
