Why I Started Paranormal Distributions

Every dataset has a ghost.

Not literally but there’s always something in the data that doesn’t behave the way the textbook says it should. A distribution that skews when it shouldn’t. A metric that moves without an obvious cause. A cohort that refuses to fit the model.

After a few years working in analytics, I’ve come to believe that the most interesting part of this job isn’t the clean, well-labelled insight. It’s the moment just before, when something looks wrong, and you don’t know yet whether it’s noise, a bug, or a signal that changes everything.

This blog is about those moments.

I work as a Business Analyst at Swiggy, mostly on Dineout, the restaurant dining side of the business. My day-to-day spans session attribution pipelines, user segmentation, campaign measurement, and a lot of SQL. The problems are rarely textbook. The data is almost never clean. And the distributions are, more often than not, a little paranormal.

I’m writing here because I want a place to think out loud about the craft of analytics. Not just the tools, but the reasoning. How you decide what to measure. How you know when a model is lying to you. How you explain a counterintuitive result to a stakeholder who expected something simpler.

No promises on frequency. Just writing when something is worth writing about.

Welcome to Paranormal Distributions.