A measure is a data value whose quantity supports meaningful numerical operations, such as aggregation, averaging, median, minimum, or maximum. It answers “how much?” rather than describing what something is. section 3.2
Classify values by asking whether numerical operations would produce a useful result:
- A taste score is a measure: averaging scores or finding their minimum and maximum tells us something useful. section 3.2
- Fruit type is a dimension because adding or averaging fruit names is meaningless. section 3.2
- An ID may look numeric but remains a dimension when aggregation would have no meaningful interpretation. section 3.2
- Dates are typically descriptive dimensions, even when minimum and maximum dates can technically be calculated. section 3.2
Fruit scores
A fruit table can pair the dimension “fruit type” with the measure “taste score.” Grouping rows by fruit type and averaging the scores produces a comparison of typical tastiness; summing arbitrary fruit IDs would produce only numerical noise. section 3.2 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 fruit table to see measures and dimensions stored as columns beside individual observations. section 3.5.1
