Data Science Articles That Bombed This Year

Round up of this year’s least well-read articles

Adam Ross Nelson
4 min readDec 20, 2023


If I was a conspiracy theorist I’d say the computers might be behind helping to reduce the readership of Let AI Tell You About It’s Flaw (with less than a few dozen reads). This article isn’t about conspiracy. Instead it is a way to learn something from the past year as we get ready to move into 2024.

Negative Case Analysis

When you want to know how something works, consider studying the things that didn’t work. Doing so, is a form of negative case analysis.

Below, I’m introducing an annual series on articles that I thought might get well-read but that just fizzled.

Negative case analysis is an approach to research that compares and contrasts what works and what doesn’t. When done well this strategy can find deep insights not available from other methods.

For example, when studying “successful businesses,” an exploration of failed businesses may provide valuable lessons. These cases often highlight pitfalls, inefficient strategies, and mismanagement issues. These insights can reveal a clearer image or roadmap of what to avoid.

When developing a theoretical framework, or logic model, this kind of investigation helps discover how processes work in high resolution and with nuance.

Negative Case Analysis in Data Science

A data scientist can use negative case analysis to study, monitor, and improve model performance over time. For example, at an e-commerce shopping platform, a data scientist’s model might have predicted a shopper to make a purchase. But the shopper didn’t pay for the items in their cart.

The negative case analysis would involve a careful review of that false positive, along with other false positives, in order to see what went awry. The ultimate goal will be to take this lessons and feed them back into the data science process for an update to the predictive models.

Articles That Few Read . . .



Adam Ross Nelson

Ask about my free career course. I mentor new (💫) and aspiring data scientists enter (🚪) and level up (📈) in the field.