munging

Munging is a term used in information technology to describe the process of transforming and cleaning data to make it suitable for analysis or processing. It is often contrasted with broader data processing in that it emphasizes programmatic manipulation to correct, reformat, or consolidate data from disparate sources. The practice is common in data science, data analytics, and software development when raw data is inconsistent, incomplete, or in a nonstandard format.

Common tasks in data munging include parsing, cleaning (handling missing or erroneous values), normalization and standardization of formats, re-encoding characters, type coercion, deduplication, and restructuring data into a consistent schema. Munging can involve simple string operations on text data, such as splitting, joining, or replacing substrings, as well as more complex transformations like pivoting, aggregation, or feature engineering.

In programming contexts, munging may also refer to deliberately altering data to test system resilience or to obfuscate information, though this usage is informal and highly context-dependent. The term is widely used in data science and software development communities and is generally preferred to describe a broader set of data preparation activities alongside more formal terms such as data cleaning or data wrangling.

Related concepts include data cleaning, data wrangling, data transformation, and ETL (extract, transform, load). The exact origin of the term is informal, with its popularity growing alongside the rise of data-driven workflows in the 2000s.