4.3.1. The Analyze Feature, Grouping, Binning, and Clustering
💡 First Principle: Before you can explain a pattern, you often need to reshape how data is grouped: combine many small categories into meaningful groups, turn continuous numbers into ranges, or let an algorithm find natural clusters. The Analyze feature then searches those dimensions for the biggest contributors to a change.
Analyze (right-click a data point > Analyze):
- Explain the increase / Explain the decrease: for a data point on a time-based column or line chart, Power BI compares it to the previous point and searches other fields in the model for the categories that contributed most to the change, presenting results as small visuals you can add to the report.
- Find where this distribution is different: for a categorical data point, Power BI looks for fields where the distribution for this point differs most from the overall distribution.
These work best on Import models with additive measures; they don't support every measure type or DirectQuery scenario.
Groups (right-click a categorical field or select data points > Group data): combine values into a new group field, for example grouping 40 countries into regions or several small products into "Other." Groups appear as new fields with a group icon and can be edited.
Bins (right-click a numeric or date field > New group > Bin type): divide continuous values into equal-sized ranges, by bin size (every 10 years of age, every $100 of order value) or number of bins. Binning creates a categorical field you can put on a histogram-style column chart.
Clustering (scatter chart > More options > Automatically find clusters): Power BI runs a clustering algorithm on the plotted measures and creates a new cluster field assigning each item to a segment. You can set the number of clusters or let Power BI choose, and edit the field later. It's useful for customer segmentation and spotting groups that behave differently.
⚠️ Exam Trap: "Create age ranges of 10 years for a histogram" is binning (bin size 10) on the Age column, not groups or a slicer. "Combine several small countries into a 'Rest of world' category" is grouping. "Find natural segments of customers based on frequency and spend" is clustering on a scatter chart.
Reflection Question: Explain the increase shows that Product Line A drove most of a revenue jump in Q2. What additional checks would you do before reporting Product Line A as the cause, and which visual could help explore further?
