Chapter 16 Glossary
The glossary provides a quick reference to concepts used throughout the book. Detailed definitions and applications appear in the chapter where each term is introduced.
Accountability: The obligation to complete agreed work, communicate progress and problems, and accept review.
Analytical plan: A documented explanation of the question, data, methods, validation, outputs, assumptions, and decision criteria.
Analytical question: A question that identifies what will be described, compared, explained, predicted, or projected using data.
Audience: The people for whom a communication is designed and who will interpret or use it.
Baseline: A simple reference method or condition used for comparison.
Benchmark: A reference value or standard used to evaluate performance or position.
Call to action: A clear statement of the response or next step an audience is asked to consider.
Churn rate: The proportion of eligible customers lost during a defined period.
Citation: A formal acknowledgement identifying the source of data, ideas, methods, or other material.
Client: The person or organization that commissions, sponsors, receives, or uses project work.
Clustering: An unsupervised method that groups observations according to a defined measure of similarity.
Codebook: A structured description of variables and coded values.
Comparability: The degree to which values from different sources can be meaningfully compared.
Confidentiality: The obligation to prevent unauthorized disclosure of information.
Customer acquisition cost: The attributable cost of acquiring a new customer under a stated allocation rule.
Customer lifetime value: The estimated economic value produced by a customer over the relationship under a stated model.
Data dictionary: A structured reference documenting the meaning, format, values, source, quality, and use of data elements.
Data leakage: Use of information during model development that would not legitimately be available when a prediction is made.
Data lineage: The traceable path from a source through transformations to an output.
Data stewardship: Responsible management of data throughout its lifecycle.
Data-use agreement: A formal agreement defining permitted data, users, purposes, safeguards, disclosure, and retention.
Deliverable: A defined product that must satisfy stated content and quality expectations.
Dependency: A relationship in which one task or input must be completed before another can proceed.
Derived variable: A variable created from source fields using a documented rule.
Discount rate: The rate used to convert future amounts into present value.
Executive summary: A concise, self-contained account of the problem, evidence, conclusions, recommendations, and material limitations.
Exploratory data analysis: A structured investigation of data quality, distributions, relationships, limitations, and analytical possibilities.
Forecast: An estimate of a future value based primarily on patterns learned from observed data.
Handoff: The organized transfer of files, documentation, context, and usage guidance.
Implementation question: A question about how an action will be carried out, monitored, and adjusted.
Interim review: A structured examination of progress, evidence, risk, and remaining work before final delivery.
Iteration: A deliberate return to an earlier stage so that work can be revised using new evidence.
Metadata: Information describing a dataset’s content, structure, collection, quality, ownership, and use.
Milestone: A meaningful checkpoint indicating that a stage is complete or ready for review.
Missingness: The amount, pattern, and process through which expected values are absent.
Practical significance: The degree to which a result is consequential enough to matter for a decision.
Primary data: Data collected directly for the purpose currently being investigated.
Privacy: Rights and expectations associated with information about individuals.
Projection: A conditional estimate of a future quantity under stated assumptions.
Provenance: The documented origin, context, ownership, version, and transformation history of data.
Recommendation: A proposed action responding to an established problem or opportunity and supported by evidence.
Reproducibility: The ability to recreate results from documented data, code, settings, transformations, and decisions.
Retention rate: The proportion of eligible customers who remain active from one defined period to the next.
Risk register: A maintained record of possible events that could harm the project and the planned response.
Scenario: A coherent combination of assumptions used to examine one possible outcome.
Scope: The agreed boundary of the questions, data, methods, periods, and outputs included in a project.
Sensitivity analysis: An examination of how results change when important assumptions or inputs change.
SMART objective: An implementation objective that is specific, measurable, achievable, relevant, and time-bound.
Storyboard: A preliminary sequence of messages and evidence used to design a coherent communication.
Team charter: A documented operating agreement governing how a project team will work together.
Thematic cohesion: Alignment of questions, evidence, language, conclusions, and recommendations around a central purpose.
Unit of analysis: The entity represented by one observation in an analysis.
Utilization rate: The number of events or services used relative to a defined population and period.
Validation: Evaluation that estimates how a method performs beyond the data used to fit or select it.
Visual encoding: The use of visual properties such as position, length, colour, or shape to represent data.