CA ANZ Data Analytics and Insights flashcards

34 cards. Read the term and the definition here. The deck at the top of this page is optional.

Data completeness rate

The percentage of required field values that are present and non-null, calculated as non-null values divided by total records, multiplied by one hundred.

Referential integrity

A condition where every foreign key value in a child table matches an existing primary key in the parent table, so no orphan records exist.

Fuzzy matching

A record-linkage technique that scores similarity between strings using edit distance or phonetic keys, used to detect near-duplicate names and addresses that exact matching misses.

Data lineage

A documented trace from a reported figure back through each transformation to its source system and field, supporting reproducibility and audit of reported numbers.

Benford's Law first digit

For naturally occurring magnitudes spanning several orders, the expected first-digit frequency is log10 of one plus one over the digit, giving about 30.1 percent for digit one.

Descriptive versus predictive analytics

Descriptive analytics summarises what has occurred in historical data, while predictive analytics estimates the likelihood of future outcomes using models fitted to that history.

Coefficient of determination

The proportion of variance in the dependent variable explained by the regression model, ranging from zero to one; a high value does not establish causation.

Interquartile range outlier rule

A value lying more than 1.5 interquartile ranges below the first quartile or above the third quartile is flagged as a mild outlier in a box-plot analysis.

Sampling risk

The risk that a conclusion drawn from a sample differs from the conclusion that testing the entire population would have produced, reduced by increasing sample size.

SQL inner join

A join returning only rows with matching key values in both tables; unmatched rows from either table are excluded from the result set.

Chart selection rule

Comparison across categories uses bars, trends over time use lines, part-to-whole uses stacked bars or treemaps, distributions use histograms, and relationships use scatter plots.

Truncated axis distortion

Starting a bar chart's value axis above zero exaggerates differences in bar length, misleading readers about the relative magnitude of the categories shown.

Data-ink ratio

The proportion of a chart's ink that encodes actual data, maximised by removing gridlines, borders, shadows, and three-dimensional effects that carry no information.

Simpson's paradox

A trend that appears in aggregated data but reverses when the same data is split into its constituent groups, so segment-level reporting is required.

Insight statement components

A well-formed insight names the finding, its magnitude, the population or period it applies to, and the decision implication for the reader.

Dual-axis chart risk

Plotting two measures on separate scales can manufacture apparent correlation; such charts require explicit axis labels and a justified reason for combining the series.

Audit risk model

Audit risk equals inherent risk multiplied by control risk multiplied by detection risk; data analytics mainly reduces detection risk by broadening testing coverage.

Journal entry testing criteria

Selection rules targeting entries posted outside business hours, in round amounts, by unusual users, to seldom-used accounts, or after period close.

Three-way match

A control comparing purchase order, goods receipt, and supplier invoice for quantity and price agreement, with mismatches, duplicates, and missing receipts flagged as exceptions.

IPE completeness testing

Testing and documenting that information produced by the entity is complete and accurate before relying on analytics results as audit evidence.

Exception disposition

The documented resolution of each analytics exception as valid, invalid, or escalated; unexplained exceptions cannot be disregarded in forming a conclusion.

Volume and price variance

A decomposition of the actual-versus-budget gap into a quantity effect, the quantity change times standard price, and a price effect, the price change times actual quantity.

Break-even units

The sales quantity at which total contribution equals fixed costs, calculated as fixed costs divided by contribution margin per unit.

Scenario versus sensitivity analysis

Sensitivity analysis varies one input at a time, while scenario analysis changes several correlated inputs together to model a coherent alternative future state.

Rolling forecast

A forecast continuously extended by adding a new period as each one closes, replacing the fixed annual budget horizon and reducing stale assumptions.

KPI definition standard

A key performance indicator requires a documented formula, data source, accountable owner, target, and review cadence so results remain comparable across periods.

APES 110 fundamental principles

Integrity, objectivity, professional competence and due care, confidentiality, and professional behaviour, which bind members performing data analytics work.

Data minimisation principle

Collecting and retaining only the personal information necessary for a stated purpose, reducing privacy exposure and the scope of any data breach.

Algorithmic bias

Systematically unfair model outputs arising from training data that under-represents a group or encodes historical discrimination, detected by subgroup performance testing.

Analytics accountability

Responsibility for a conclusion drawn from analytics rests with the professional who signs it, irrespective of who built the underlying model or tool.

RPA versus machine learning

Robotic process automation executes deterministic rules on structured inputs, whereas machine learning generalises patterns from labelled examples to handle judgement-like tasks.

Process mining

Analysis of system event logs to reconstruct actual process flows, revealing bottlenecks, rework loops, and deviations from the designed control path.

Continuous auditing

Automated recurring tests run against live or near-live data, distinct from continuous monitoring, which is management's own real-time control process.

Model risk management

A framework covering model inventory, independent validation, ongoing performance monitoring, and documented retirement of superseded models to control reliance risk.