data-profiling

Structural Reconnaissance

Table-Level Inventory

Start every new dataset encounter by answering these questions:

• What is the total row and column count?
• What does each row represent (the grain)?
• Which column or columns uniquely identify a row?
• When was the most recent data loaded?
• What is the earliest date in the dataset?

Classifying Columns by Role

Assign every column to one of these categories:

• Key: Primary keys, foreign keys, entity identifiers
• Attribute: Categorical fields used for grouping or filtering (region, status, plan_type)
• Measure: Numeric fields intended for aggregation (revenue, duration, score)