The data science hierarchy of needs (reread)

About 18 months ago, I read for the first time about “the data science hierarchy of needs”, as presented by Monica Rogati (with links to Maslows idea +/- 70 years ago). I still use the idea behind it regularly (in a simplified form) in discussions with clients. Last week, I read the original article again, to check if it is still relevant.

My conclusion: it still makes much sense. Data in the lower part, Business Intelligence in the middle part and more sophisticated Analytics and AI in the top part. Learn to walk before starting to run.

And even nuances I usually make, are already present in the paper in some way.
Such as that things don’t need to be really perfect (“The data science hierarchy of needs is not an excuse to build disconnected, over-engineered infrastructure for a year.”). Don’t think you can’t have Business Intelligence when you don’t have perfect data, you will also learn to improve your data after you gain more insight in them. But of course, some minimum quality level is needed.

Another nuance already present in the article is that you shouldn’t see it too much as a layer-by-layer approach. Eg more advanced Business Analytics should not always wait for the perfect BI solution to be in place.
On the other hand, don’t tackle technologies like AI as a hype, just for proving that you do something with it. You will fail.
If you’re going for a first project with AI, succeeding will be more important for future success than trying to tackle the most profitable idea the first time. That’s an interesting topic for a future blog.

Next reread of Monica’s article in 18 months…