Big data and data analytics: formula sheet

Key formulas

Source classification

Internal = generated inside the organisation; External = obtained from outside it

Ask who created the data and where it was captured. Bought data is external even once stored in your system.

Structured data

Structured = fixed format, rows and columns, predefined fields

Examples: ledgers, inventory tables, payroll records.

Unstructured data

Unstructured = no predefined format or data model

Examples: emails body text, images, video, audio, social media posts.

Semi-structured data

Semi-structured = tags or markers give partial organisation, no rigid table

Examples: XML, JSON, emails with header fields, web pages.

Descriptive analytics

Question: What happened?

Summarises past data. Typical outputs: reports, dashboards, averages, totals.

Diagnostic analytics

Question: Why did it happen?

Finds causes. Typical methods: drill-down, comparison, correlation, variance investigation.

Predictive analytics

Question: What is likely to happen?

Estimates future outcomes using past data and models. Gives likelihoods, not certainty.

Prescriptive analytics

Question: What should we do?

Recommends the best action, often using optimisation or simulation. Needs the most data and skill.

Data analytics vs big data

Big data = the data; data analytics = the analysis of data

Analytics can be applied to data of any size.

Quick revision

• The five Vs are volume, velocity, variety, veracity and value.
• Volume is the amount of data; velocity is the speed it arrives and must be processed.
• Variety means many formats; veracity means how accurate and reliable the data is.
• Value means the data is only worth having if it leads to better decisions.
• Structured data fits fixed fields, such as a database; unstructured data, such as emails or video, does not.
• Descriptive analytics shows what happened; diagnostic explains why.
• Predictive analytics estimates what is likely to happen; prescriptive suggests what to do.
• Big data brings better insight and faster decisions, but also cost, poor-quality data and privacy risks.
• Correlation found in data does not prove one thing causes another.
• Cloud computing gives scalable storage and processing without owning all the hardware, but raises security and dependence concerns.
• Always tie your answer to the scenario business, not just the textbook definition.

Common mistakes

• Confusing velocity with volume. Fix: Volume is how much data. Velocity is how fast it arrives and must be processed. Use the words 'amount' and 'speed' to separate them.
• Listing the five Vs with no example. Fix: Add one line of application to each V using details from the scenario. Application earns the marks.
• Calling all big data unstructured. Fix: Big data includes all three types. Classify each item on its own format.
• Treating an email as purely unstructured. Fix: Say an email is semi-structured if the question considers its fields. Unstructured fits the body text alone.
• Treating data analytics and big data as the same thing. Fix: Say big data is the data (volume, velocity, variety and so on) and analytics is the analysis of it. Analytics works on small data too.
• Calling a forecast prescriptive. Fix: Predictive says what will probably happen. Prescriptive goes further and recommends an action.
• Listing generic points with no link to the scenario. Fix: Name the business and its data in every point, for example the retailer's loyalty card data.
• Writing only benefits or only risks. Fix: Check the verb. If it says discuss or evaluate, cover both sides and conclude.
• Saying cloud computing and big data are the same thing. Fix: Big data is the data itself. Cloud computing is a way of delivering storage and processing. Cloud can host big data, but they are different ideas.
• Claiming AI will replace the management accountant. Fix: Say AI automates routine tasks, while the accountant interprets output, challenges assumptions, advises managers and checks data quality.

Exam tips

• Learn the five Vs as a fixed list and be ready to define each in one line. Objective test questions often give a short description and ask which V it shows.
• In written answers, always tie each V to the scenario. A bare definition scores little.
• Watch for veracity being tested through 'unreliable' or 'inaccurate' wording, and value through 'benefit exceeds cost'.
• Use big data points in wider questions on performance measurement, forecasting and information systems, not only in direct big data questions.
• If you are asked for characteristics and only three are named in your notes, still add veracity and value when the question says five or does not limit you.
• Read each data item and decide source and type separately. Questions often test both at once.
• In Section C, always tie the data to the business in the scenario. Generic lists score poorly.
• Watch for options where one is internal but the wrong type. Only one option matches both.