Fixed vs Rolling Time Windows? - Issue 175

When to use x-day or rolling time window for retention, time series, or other analysis

You're starting a new role as an analyst, and the first project you’ve received is to build a year-to-date report of new signups and a year-over-year rolling average of sales. You know how to pull the data, and you already googled “year-to-date”, but you’re still confused about what exactly the year-over-year rolling average would mean because the 12-week Data Science Immersive Bootcamp for $16,450 at General Assembly didn’t quite cover that.

I have got your back. Today, I will try to bridge the gap between what you have been taught and what is expected of you to know to keep your analyst hat on. I’ll share my approach on how to figure out when to use fixed (also known as calendar) vs rolling (also known as moving) time windows. I will explain what “year-to-date” means, and how to determine the right time window for retention analysis.