Grain
The meaning of one row: order, order line, customer-month or another explicitly defined unit.
Explore & practiseThe practice & career studioTurn what you know
Turn what you know
into work you can explain.
Investigate a mistake. Solve a small challenge. Build a case you can defend. Original resources that help you move from reading to doing.
Find your next activity ↓A useful place to begin
Choose a small win.
Explore your companion library
Download companion pack ↓Practical glossary
30 resources available. No sign-up required.
1130 resources
Practical glossary
Understand the language through examples.
Primary key
A field or combination that uniquely identifies a record and is not missing.
Explore & practiseForeign key
A reference to a key in another table; validate unmatched values before trusting the join.
Explore & practiseMissing value
A value that is unavailable or unknown; it is not automatically zero or an empty category.
Explore & practiseDenominator
The population or exposure count used as the bottom of a ratio.
Explore & practiseBaseline
A simple reference method against which a more complex approach is compared.
Explore & practiseData leakage
Information unavailable at the prediction time or improperly crossing evaluation boundaries.
Explore & practiseValidation
Explicit checks that data or outputs meet a stated contract.
Explore & practiseReconciliation
Comparing totals or records between sources and transformations, with explained differences.
Explore & practiseAggregation
Combining records into a coarser output grain using defined functions.
Explore & practiseWeighted rate
A rate computed from total numerators and denominators or equivalent denominator weights.
Explore & practisePercentage point
An absolute difference between two percentage values.
Explore & practiseAttribution
A rule assigning an observed conversion to a channel or interaction.
Explore & practiseCausality
A claim that changing one factor changes an outcome, requiring a suitable design and assumptions.
Explore & practiseHoldout
Data reserved from model fitting and selection for a later evaluation.
Explore & practiseOverfitting
Adapting too closely to training noise or reused evaluation feedback.
Explore & practisePrecision
Among predicted positives, the fraction that are true positives; undefined if none are predicted positive.
Explore & practiseRecall
Among actual positives, the fraction correctly predicted positive; undefined when no actual positives exist.
Explore & practiseDrift
A change in data or relationships relative to a reference; its operational meaning needs investigation.
Explore & practiseIdempotency
Rerunning an operation with the same inputs does not create unintended additional effects.
Explore & practiseProvenance
A record of sources and transformations that explains where an output came from.
Explore & practiseFile hash
A digest identifying exact file bytes; it does not establish truth or permission.
Explore & practiseOutlier
An unusual observation relative to a specified distribution or rule.
Explore & practiseImputation
Replacing missing values using an explicit method and assumptions.
Explore & practiseNormalization
A term with different meanings in data cleaning, numerical scaling and relational design.
Explore & practiseUncertainty
Limits on what can be concluded from available evidence and assumptions.
Explore & practiseRubric
Criteria with defined evidence and levels used to review work consistently.
Explore & practisePortfolio
A collection of your own inspectable work with attribution and honest context.
Explore & practiseReproducibility
The ability to rerun a documented method on identified inputs and obtain explained results.
Explore & practiseAccessibility
Designing content and interactions so people with different abilities can use them.
Explore & practiseResume guidance and examples are for learning and preparation only. They do not guarantee interviews, a job offer or employment.
Public registration, payments and new paid enrolment remain disabled. These free resources do not grant access to protected course modules.
Learner dashboard ↗ · Ask a question ↗