Is Tabular Data Boring?

Struggling to get people excited about tabular data when compared to the buzz surrounding AI for images, voice, text, video, and other unstructured data? The odds are stacked against tabular data, but it doesn’t have to be this way!

Humans are wired to process the sights and sounds around us quickly – it’s in our evolutionary DNA. We’re visual learners, with the ability to process images in as little as 13 milliseconds – 60,000 times faster than text. And research shows that images can dramatically improve user engagement in social media. It’s no wonder text-to-image generators and ChatGPT have gone viral.

On the other hand, interesting tabular data is usually behind firewalls, deep in the bowels of businesses and organizations. Data is rarely exposed to people enough to become intuitive in their daily lives. We need to exert extra effort to overcome the inertia that our brains are wired with when dealing with complexity.

Common data science practices may be part of the problem too. Much of our feature engineering activity focuses on transforming data in single tables, which can lead to the dreaded “yawn factor.” Let’s face it, one-hot-encoding isn’t the most exciting or challenging task you will ever encounter. 

But that doesn’t mean that tabular data isn’t interesting! When data scientists use their curiosity and imagination, they can find insights well beyond standard RFM (recency, frequency, monetary) signals. With event sequences, complex human and market behaviors, similarities of attributes, seasonality, and more at play, there’s plenty of opportunity to derive new signal types and uncover amazing insights.

So, if you’re working on tabular data, use your human creativity to make it interesting. What is the most exciting analysis you’ve done with tabular data?