The Keys to Effective Data Science Projects – Part 8: Operationalize

readWe’re in part eight on our journey through the series of the Keys to Effective Data Science Projects –“Operationalization” – a term only a marketer could love. It really just means “people using your solution”. And it’s this part of the process that is quite possibly the most complicated, and usually the one done with the least thought.

In days gone by, the Data Scientist used R or Python for analysis or prediction and created a visualization for their audience. Or perhaps they returned a few rows of data, something someone could put in an Excel spreadsheet.

And that’s not a bad thing – in fact, visualizations are often the end-result of what you’re working on. But you have more options than just these.

That’s the first key: Way back during the very first design phase, have a discussion about the various ways that people might interact with your solution.

And that’s the second key: You don’t have to use the result in just one way. Think through other ways your hard work can be used. You can expose an API, create a visual, output some data, stream some data, trigger another process and more.

There are a few general ways you can allow people to interact with something you’ve built, and most of the time the way you use it is based on what the output is. Let’s take a moment and suss that out a bit.

If you’re creating a clustering algorithm, you might surface the results in a panel on a web page – “customers who bought this item also liked these items!” or some such. So “Operationalizing” this result would be to expose an Application Programming Interface (API) that is called whenever a customer buys something on the website.

But should you consider a few more options? Could that API be surfaced at a brick-and-mortar store on a receipt with a discount on the grouped items? Could you store that prediction to see if the customers really did buy those other items. What other things can you think about that would benefit from your solution?

Or perhaps you’re doing a prediction of which medicines to recommend to a particular patient. While that would be a great thing to have available at the doctor’s fingertips while they interact with the patient, would it also be useful to the drug manufacturer to know which patients would benefit best from their pharmaceuticals?

The key is not to pick one mechanism over another – it’s to consider, right from the start, all the ways the solution can be used.



2 thoughts on “The Keys to Effective Data Science Projects – Part 8: Operationalize

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.