DS Nuggets — Why every data scientist needs to know UI design
The next big thing is the one that makes the last big thing usable. —Blake Ross, Co-creator of Mozilla Firefox
Thou shall train, thou shall test, thou shall deploy but who shall use? The majority of data science literature focuses on shimmery algorithms, innovative evaluation methods and system design. I wont deny these are all essential steps in the sequence of making machine learning models operational aka useful.
Useful is not always usable!
If your model populates values into a database, then it’s probably useful already. But if your model demonstrates insights directly to an end user, what you need is a UI.
You might be doing it already
One of the first apps that data scientists learn to use is Excel. I can’t remember the sheer number of times I’ve presented my analysis in a workbook — pivot tables, charts, slicers, wow! As of writing this post, Excel has started to support Python code in cells.
I won’t deny it helped me furnish reports quickly when I’ve been pressed for time.
As my analysis got more complex, so did the stakeholders’ requests.
You know what’s limiting about a workbook analysis? — You have to send or share the file. It…
