Introducing rolling averages
Today, we’re annoucing the ability to do rolling averages on any of your time
series data. Many of you might have already noticed and played with the
feature. You can find it under the “Analysis” menu option right above the
What are rolling averages?
Given a series of numbers and a fixed subset size, the moving average can be
obtained by first taking the average of the first subset. The fixed subset
size is then shifted forward, creating a new subset of numbers, which is
averaged. This process is repeated over the entire data series. The plot line
connecting all the (fixed) averages is the moving average.
For example, if you the number of page views you received were 1, 2, 10, 12,
13, 100, 11 for each day and we did a 3 day rolling average. The first one
would be: 1, the next: 1.5, the next: 4.3 ((10 + 2 + 1) / 3), the next: 8.
Hopefully you get the point.
Why are rolling averages even useful?
Rolling averages are incredibly useful for understanding your traction. A 7-10
day over day rolling average over a large enough time period that trends
upwards might indicative of impressive growth depending on the rate.
Rolling averages smooth out spikes you may get from press and blogs talking
about you so you can see how you are really growing. In the two images we
posted you’ll notice one has troughs from the weekend as well as spikes that
are smoothed out in the next graph.
Try it out, if you have a 30 day rolling average that’s trending upwards you
might be on to something and thankfully with Mixpanel you can do this on any
To give you an example of how we use it personally. We deem people playing our
video to be indicative of growth and awareness in terms of marketing. This is
an actionable metric of someone expressing interest and learning about our
product so a 7 day rolling average over 30-60 days will help support that
Mixpanel’s marketing efforts are working.