VCE Applied Computing Data Analytics flashcards

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

Data integrity

The accuracy, completeness, consistency, timeliness and validity of data across its life cycle. Weak integrity at any stage undermines every later analysis and decision made from the dataset.

Validation versus verification

Validation checks that entered data meets predefined rules such as type, range and format. Verification confirms the data matches its original source, for example by double entry or cross-checking documents.

Referential integrity

A rule requiring every foreign key value in a table to match an existing primary key in a related table. Records with unmatched keys are orphans and break reliable joins.

Data life cycle

The stages data passes through: collection, storage, preparation, analysis, visualisation and communication, then archiving or disposal. Each stage can introduce error or reduce data quality.

Metadata

Data that describes other data, such as field names, units, formats, collection dates and provenance. Without metadata, users cannot correctly interpret, compare or reuse a dataset.

Data steward

The role responsible for maintaining data quality, standards and definitions within a domain. Stewards differ from owners, who are accountable, and custodians, who manage storage and security.

Primary data

Data collected firsthand for the current purpose, such as surveys, interviews, sensors or experiments. The collector controls the method but usually bears higher cost and time.

Secondary data

Data originally collected by someone else for another purpose, such as census files or government open data. It is cheaper and faster but may not match the current need.

Stratified sampling

A method that divides the population into defined strata, then randomly samples within each stratum. It guarantees representation of each group and reduces sampling error compared with simple random sampling.

Convenience sampling

Selecting participants who are easiest to reach, such as volunteers or bystanders. It is fast and cheap but usually non-representative, so findings cannot be generalised to the population.

Response bias

A systematic distortion caused when people who respond differ from those who do not, or answer inaccurately due to question wording or social pressure. It skews results in a consistent direction.

Leading question

A survey question phrased so it suggests or encourages a particular answer. It biases responses and is a design fault corrected by neutral, balanced wording.

Quantitative data

Numeric data representing counts or measurements, such as age, temperature or sales volume. It supports arithmetic and statistical analysis, unlike descriptive qualitative data.

Data cleaning

Detecting and correcting errors such as duplicates, inconsistent formats, typos and invalid values. Cleaning improves accuracy and consistency before analysis and should be documented.

Imputation

Replacing missing values with estimated ones, such as the mean, median, mode or a predicted value. It preserves sample size but introduces uncertainty that should be reported.

Outlier

A value markedly different from the rest of the dataset. It may be a genuine extreme observation or a recording error, so it must be investigated before removal.

Interquartile range rule

A method flagging values below Q1 minus 1.5 times the IQR or above Q3 plus 1.5 times the IQR as potential outliers. It is robust because quartiles resist extreme values.

Data wrangling

Transforming raw data into an analysable form through filtering, sorting, joining tables, aggregating, pivoting and deriving new fields. Each transformation should be logged for reproducibility.

Min-max normalisation

Rescaling numeric values to a 0 to 1 range using (x minus minimum) divided by (maximum minus minimum). It preserves relative order but is sensitive to extreme values.

Joining tables

Combining two tables using a shared key field. An inner join keeps only matching records, while outer joins also retain unmatched records from one or both tables.

Scatter plot

A chart plotting one numeric variable against another to reveal relationships, clusters and outliers. It suits correlation analysis but does not by itself establish causation.

Truncated axis

A graph axis that does not start at zero, exaggerating differences between bars. It is acceptable for line charts of trends but misleading for comparing magnitudes.

Data-ink ratio

The proportion of a chart's ink that represents actual data rather than decoration. Maximising it by removing gridlines, 3D effects and clutter improves clarity and honesty.

Pie chart suitability

Pie charts suit part-to-whole comparisons with few categories summing to 100 percent. They are poor for comparing many categories or showing change over time; bar charts are better.

Correlation versus causation

A statistical association between two variables does not prove one causes the other. A confounding variable may drive both, so causal claims need controlled or randomised evidence.

Median

The middle value when data is ordered. It is a robust measure of centre that resists distortion by outliers, unlike the mean, which extreme values pull sharply.

Standard deviation

A measure of the spread of values around the mean. A small value indicates data clustered near the mean; a large value indicates wide dispersion.

Confounding variable

An unmeasured factor that influences both the supposed cause and the observed outcome, creating a spurious association. It must be controlled to support causal interpretation.

Benchmark

A reference point such as a target, historical baseline or peer result used to judge whether an observed value is good, bad or unusual. Numbers lack meaning without one.

De-identification

Removing or altering identifiers so individuals cannot reasonably be identified. Risk remains when combined quasi-identifiers such as age, postcode and birth date uniquely isolate a person.

Informed consent

Agreement to data collection given after being told the purpose, what is collected, who receives it and the right to withdraw. Consent obtained by deception is invalid.

Least privilege

A security principle granting users only the access needed for their role. It limits the damage from compromised accounts and supports privacy obligations for personal information.

Critical path

The longest sequence of dependent project tasks, which determines the minimum project duration. Any delay on the critical path delays the whole project.

Scope creep

Uncontrolled expansion of project requirements beyond the agreed scope. It is managed through formal change requests assessed for impact on time, cost and quality.

Milestone

A zero-duration checkpoint marking completion of a key deliverable or phase. Milestones anchor progress reporting but consume no effort themselves.