The concept of data granularity makes perfect sense… but as data newbie it can be a little confusing when we start talking about “high” vs “low” granularity, or “high” vs “low level of granularity.”
What is granularity?
In simple terms granularity of a dataset is how detailed and specific one record or row of data is.
I like to think of granularity like sand:
- High granularity = fine grains of sand = very detailed, specific data
- Low granularity = coarse grains, or even clumps or an entire beach of sand = summarized, aggregated data
We can change the granularity of a dataset in two main ways:
1. Adding more details by having more categories or fields of data thereby increasing granularity
2. Aggregating data summarising by totals or averages, thereby decreasing granularity.
For example:
High granularity = each record shows an order transaction with details of the date, store, product and sale value of the order
Medium granularity = each record shows the daily sales per store
Low granularity = each record shows the total sales for each store
As the granularity decrease the dataset becomes smaller, with fewer rows and often fewer fields of data.
Why the terminology is confusing
The confusion in terminology lies in when we talk about the level of granularity.
High granularity will show data at a low level of granularity.
And
Low granularity will show data at a high level of granularity.
Confusing right?
I find it’s easier to think about it like using a camera lens.
Zooming in we get a close up or low view – seeing tiny details like each grain of sand.
Zooming out we get a higher, broader/wider view - seeing the whole beach.
So:
Lower view (level of) = closer view which is specific = higher granularity
Higher view (level of) = wider view which is summarised = low granularity
I hope that’s made it a bit clearer if you’re a fellow data newbie!
In short, granularity is just your data’s zoom level. Zoom in for detail, zoom out for summaries. Knowing how to switch between the two is what makes great analysis possible.