SECAI Core · phase 3 of 15 · noun

Data cleaning

The process of ensuring that the data used in analysis is of a high-enough quality that it gives decision makers confidence in findings and the practice of transforming non-conforming data into acceptable shape.

The Explain card

Plain English
Data cleaning makes sure data is good enough to trust: fixing errors, filling or removing bad values, and reshaping non-conforming records into an acceptable form.
Example
Before training a fraud model, analysts strip out test transactions, correct mistyped currency codes and drop records with impossible dates.
Why it matters
Dirty data produces unreliable models, but cleaning is also a security step. It is where poisoned outliers, injected instructions and leaked secrets can be caught before they are learned.
Hook
Garbage in, garbage learned; clean before you train.

Word knowledge

How the word is built, where it came from, and what it sits beside in memory.

In a sentence

Before the fraud classifier was trained, data cleaning removed test transactions and rows with impossible timestamps from the training set.

Why these words
  • data Latin data, things given, plural of datum supplies the given-facts half, so the activity is applied to records
  • cleaning Old English clæne, clear of dirt, plus -ing supplies the head noun, the activity the phrase assigns to those records
Where it came from
Origin
English computing collocation from Latin data, things given, plural of datum, and Old English clæne, clear of dirt, plus -ing
Entered the language
1980s
What changed
Data had meant things given, then facts granted in an argument, then the records a machine keeps. Cleaning had meant making a house, a conscience, or a street clear of dirt. Statistical computing of the 1980s joined the pair as the name of a pipeline step. Warehouse literature of the 1990s also wrote data cleansing, and machine-learning labs later used both names for the pass that runs on a training set before a model is fit.
How it is spelled
Pattern
data stays the Latin plural of datum, even when English treats it as a mass noun
Pattern
ea spells the long /iː/ in clean, as it does in mean and lean
Pattern
a long vowel before a single consonant does not double: cleaning, not cleanning
Pattern
-ing turns a verb into a noun of activity
Breaks the pattern
the noun phrase stays two words; the hyphen appears only when it modifies another noun
Breaks the pattern
data looks plural but computing English often treats it as a mass noun
Breaks the pattern
the related verb cleanse keeps an s the adjective never had

Spelled like

  • database
  • mean
  • lean
  • training
Broken into chunks
  • data
    • datum
    • metadata
    • database
  • clean
    • unclean
    • cleaner
    • cleanse
  • ing
    • training
    • computing
    • reasoning
What it sits beside

Same subject

Same shape

Data compounds

Cleaning compounds

  • dry cleaning
  • spring cleaning
  • house cleaning

Where it sits in the deck

Phase 3: Data Fundamentals: Types, Pipelines, and Preparation

Models are only as good as their data — understand the raw material and the engineering that shapes it before covering how it flows or breaks.

Neighbours in this phase