Big data and data analytics: formula sheet

Key formulas

Volume

Volume = how much data

Scale of data held and processed. Link to storage cost and processing capacity.

Velocity

Velocity = how fast data arrives and must be used

Real-time or near real-time data. Link to timely decisions.

Variety

Variety = structured + semi-structured + unstructured data

Different formats and sources. Link to difficulty of integration.

Veracity

Veracity = accuracy and trustworthiness of data

Poor veracity leads to poor decisions. Link to data quality controls.

Value

Value = benefit from using the data − cost of collecting, storing and analysing it

A conceptual test, not a calculation. Data is only worth keeping if it improves decisions.

Descriptive analytics

What happened? (past data, summaries)

Reports, dashboards, averages and trends. Hindsight only.

Diagnostic analytics

Why did it happen? (causes, drill-down)

Finds reasons behind results. Correlation does not prove cause.

Predictive analytics

What is likely to happen? (models, forecasts)

Uses past patterns to estimate future outcomes. Results are probabilities, not certainties.

Prescriptive analytics

What should we do? (recommended actions)

Suggests the best option given goals and constraints. Needs human judgement and oversight.

Data mining vs machine learning

Data mining = finding patterns; machine learning = learning from data to improve predictions

Data mining discovers patterns. Machine learning builds models that adapt as more data arrives.

Four value areas of big data

Decisions + Customer insight + Efficiency + New business models

A memory frame for structuring benefit answers. It is a checklist, not an official ACCA model.

Data to value chain

Data → Analytics → Insight → Action → Value

Value arises only when insight leads to action. Use it to criticise firms that collect data but do not use it.

Benefit test for any use

Benefit (revenue gained or cost saved) > Cost of data, systems, skills and risk

Use as a judgement test when asked to advise, not as a calculation unless figures are given.

Privacy vs security

Privacy = rightful use of personal data; Security = protection of data from unauthorised access

Use this distinction when a question asks for the difference. Security is needed for privacy, but is not enough on its own.

Typical data protection principles (GDPR-style)

Lawful, fair and transparent; purpose limitation; data minimisation; accuracy; storage limitation; integrity and confidentiality; accountability

Use them as a checklist against the scenario. Say 'GDPR-style' unless the scenario names a specific law.

Risk response framework

Identify risk → assess likelihood and impact → respond (treat, transfer, avoid, accept)

Links data risks to the TARA approach for recommendations.

Ethics check

Is it legal? Is it fair? Is it transparent? Would stakeholders accept it if they knew?

A quick test for ethical issues that the law may not cover.

Data governance framework elements

Policies + Standards + Roles/Ownership + Processes + Controls + Monitoring

Use as a checklist when advising on governance. Add accountability at board level.

Data quality dimensions

Accuracy, Completeness, Consistency, Timeliness, Validity, Uniqueness

Not a fixed official list. Pick the ones that matter in the scenario and explain the effect on decisions.

Data strategy building blocks

Objectives → Data needed → Technology and storage → Skills → Governance → Ethics and compliance → Costs and benefits

A logical order for an answer. Start with business objectives, not technology.

Quick revision

• Big data is data too large, fast or varied for traditional tools to process well.
• The core Vs are volume, velocity and variety; veracity (reliability) and value are often added.
• Descriptive analytics asks what happened; diagnostic asks why it happened.
• Predictive analytics asks what is likely to happen; prescriptive asks what should we do.
• Data only has value if it leads to better decisions or actions.
• Key benefits: customer insight, efficiency, better forecasting, new products and risk detection.
• Poor data quality leads to poor decisions, so question the source and reliability of data.
• Main risks: privacy breaches, cyber attack, bias, over-reliance on models and cost.
• Ethical issues include consent, fair use, transparency and not discriminating against people.
• Governance needs clear ownership, policies, access controls and board oversight.
• Always tie points to the scenario: the industry, the customers and the organisation's strategy.
• Finish data answers with a clear, justified recommendation.

Common mistakes

• Listing the Vs with textbook definitions only. Fix: Give a case-specific example and a business effect for every V you name.
• Confusing velocity with volume. Fix: Volume is how much data. Velocity is how fast it arrives and must be processed.
• Listing the four types as a definition with no link to the scenario. Fix: For each type you mention, tie it to a problem or data source in the case and state the benefit.
• Confusing diagnostic with descriptive analytics. Fix: Descriptive says what happened. Diagnostic explains why. If the question asks for causes, it is diagnostic.
• Listing generic benefits with no link to the case. Fix: Name the case's data source, customer group or process in every point.
• Describing what big data is instead of what it does for the business. Fix: Give one line of definition at most, then spend your time on benefits and application.
• Treating privacy and security as the same thing. Fix: State the difference in one line: privacy is about rightful use, security is about protection. Then give a scenario example of each.
• Writing a generic list of risks with no link to the scenario. Fix: Quote the facts given, such as the type of data or the customers, and explain the impact on this organisation.
• Treating data governance as an IT issue only. Fix: Name the board, data owners and senior management as accountable, and link governance to risk and corporate governance.
• Listing the Vs of big data or definitions without applying them. Fix: Only use definitions to support a point about the case. Every paragraph should refer to the organisation's facts.

Exam tips

• Do not stop at the list of Vs. SBL rewards application, so tie each V to the case.
• Choose the Vs that fit the scenario rather than describing all of them equally.
• Always balance benefits with limits, especially veracity, cost and data protection.
• Use professional skills: a clear structure, concise points and a firm recommendation for the board.
• If the task asks for a comparison with traditional data, use size, format, speed and tools as your headings.
• Link every type of analytics to a problem in the case. Generic definitions earn few marks.
• Use the verb in the requirement. Evaluate needs benefits and limits. Recommend needs a clear decision.
• Always include a risk and ethics point, such as privacy, bias or poor data quality.