Data aggregation combines detailed rows into summarized results. Before aggregating, identify the table’s grain: what one row represents. A transaction table, daily banking table, and account snapshot answer different questions because each has a different granularity. section 3.5.1
Use aggregation without losing track of what the numbers mean:
- Start with granular data when possible; detailed rows can be aggregated later, while summarized data cannot easily reconstruct the details. section 3.6
- Check the grain before combining tables or interpreting results, because duplicated rows or mismatched granularity can send a visualization sideways. section 3.5.1
- Match the aggregation to the question and chart: aggregated discrete data often suits bar charts, while continuous data often suits line charts. section 3.6
From individual fruits to a comparison
A detailed fruit table records each fruit type, taste score, and date eaten. Aggregating those rows produces one summarized taste score per fruit type, creating a compact table that can power a bar chart. The summary is useful precisely because its grain has changed from individual observations to fruit categories. section 3.5.1
Table 3.1 A table of data I hypothetically collected about how tasty individual fruits have been recently (view table figure)
Inspect the individual fruit records before aggregation; the repeated fruit types make the change in grain visible. section 3.5.1
