What is Data Transformation

Definition

Data transformation is the process of converting raw data into a structured, consistent, and analysis-ready form by applying operations such as cleaning, type casting, joining, aggregating, and deriving new fields so the output matches the schema and semantics that consumers expect.

« Back to Glossary Index

Reconciles inconsistent formats, units, and codes from different sources into one shared standard
Derives the calculated fields and aggregates that business questions actually require
Encodes business rules once so every downstream report applies the same logic
Improves trust by stripping out malformed, duplicate, or out-of-range records during processing

Real World Example

A bank transforms raw transaction logs by normalising merchant codes, converting all amounts to a base currency, and deriving rolling spend categories that feed its fraud-scoring models.

FAQs

What operations count as data transformation?

Cleaning, deduplication, type conversion, joining, filtering, aggregating, and deriving new calculated columns are all common transformations.

Should transformation happen before or after loading?

It depends on the pattern: ETL transforms before loading, while ELT loads raw data and transforms it inside the target platform.

How are transformation rules kept consistent?

Tools like dbt let teams define transformations as version-controlled, tested SQL models so the same logic is reused everywhere.