Innovative Tableau

Chapter 79. How to Do Anomaly Detection

I think we’ve all been there: we are sharing a meaningful story we found in the data, only to have our end users fixate on a previous peak or valley in our visualization. This often derails the conversation at hand or prevents our audience from hearing the rest of our message or reduces our chances of causing action.

One of the biggest challenges we face as data visualization practitioners is helping our end users to avoid distraction. When our end users become distracted, it makes it more challenging to communicate the story in the data and our recommended actions. Ironically, one of the reasons users become distracted is that visualizing data makes it much easier to spot points of interest. Unfortunately, just because something might pique interest, it is not always relevant to the conversation.

This chapter shares an approach to doing anomaly detection in Tableau. With anomaly detection, you’re able to focus on the data points that matter and have a statistical explanation for your end users to help avoid distracting conversations.

Using Table Calculations to Do Statistical Anomaly Detection in Tableau

To help illustrate an approach for doing anomaly detection in Tableau, we re-create the sales-by-month trend using the Sample – Superstore dataset. Note that the reference distributions show whether each data point is within one or two standard deviations from the mean and the circles are colored based on whether they are an anomaly. For ...