Topic 1 - Understanding data
1.1 Data, information and knowledge
Sources: data comes from transactions, sensors, forms and questionnaires, websites, social media, records and documents. Information comes from processing that data in reports, databases, spreadsheets and dashboards. Knowledge comes from people (experience, expertise) and from analysis of information.
How they interact: data is the input; processing (sorting, calculating, grouping, comparing) gives it context and turns it into information; interpreting information, together with experience, gives knowledge. Knowledge then decides what new data to collect, so the three form a cycle. The difference is the amount of meaning: data has none on its own, information has context, knowledge supports a decision.
How data is converted to information: collect it, check and clean it, organise it (sort, group, put into tables), process it (calculate totals, averages, comparisons), then present it with labels and context.
OCR's sample mark scheme rejects answers that repeat a column heading with no context. "Prices" is not an item of information; "the price of a large pizza is £14.00" is. For knowledge, give a conclusion such as "large is the most popular size", not another fact.
1.2 Big data
Big data is data sets so large, fast-growing and varied that traditional software (a single spreadsheet or desktop database) cannot capture, store and process them in a reasonable time. It is usually described by its characteristics, such as volume, velocity and variety (F201 covers all six Vs).
Not in F200: the evolution of big data, data preparation techniques, data mining techniques and big data infrastructure. These are F201 content.
In "explain two benefits of big data" questions, OCR gives one mark for the benefit and one for applying it to the organisation, for example "analyse past sales data across the chain (1) so each restaurant has the right number of staff at each time of year (1)". Answers not related to the scenario get no marks.
1.3 Data and file formats
A data format is the structure in which data is stored or encoded in a file, so that software knows how to read it. (OCR's sample paper asks "What is data format?" and the answer is structure.)
<price>9.00</price>; nested hierarchy. Where it is used: Exchanging data between systems, configuration files, SOAP and XML-RPC APIs. Benefits: Human and machine readable; self-describing; supports validation against a schema. Limitations: Verbose, so larger files and slower to process than JSON.{"size": "Small", "price": 9.00}. Where it is used: Web APIs (REST, JSON-RPC), web and mobile apps, NoSQL databases. Benefits: Lightweight; easy for people and programs to read; fast to parse. Limitations: No comments; fewer built-in validation features than XML.Not in F200: any format not listed above, and lossy and lossless compression.
OCR's mark scheme for "one benefit of JPEG for social media" credits: the file can be compressed so many images upload quickly; it is highly compatible so social media platforms accept it; it gives accurate colour. Each benefit needs an expansion linked to the context for the second mark.
1.4 Data types and classifications
"State two reasons why this data is structured" (sample paper): OCR credits "it is stored in an organised format" and "it is stored in tables, in columns and rows". Qualitative/quantitative is about what the data describes; structured/unstructured is about how it is organised. Keep the two pairs separate.
Topic 2 - Managing data
2.1.1 Data lifecycle management (DLM)
Data lifecycle management (DLM) is managing data through every stage of its life, from when it is created until it is destroyed. The five phases in order are:
Interactions and iterations: the phases are not strictly one way. Data in use is updated and stored again; archived data can be retrieved back into usage (for an audit or a legal request); new data created from analysis restarts the cycle. Retention rules decide when data moves from usage to archival and from archival to destruction.
Benefits of DLM: data stays accurate and available when needed; storage costs fall because old data moves to cheaper storage; security improves because each phase has controls; it supports compliance with the DPA 2018 and UK GDPR (storage limitation: do not keep personal data longer than needed); and secure destruction reduces breach risk.
The sample paper asks you to put the five phases in order. Learn the order: Creation, Storage, Usage, Archival, Destruction. Marks need the correct order, not just the names.
2.1.2 Data analytics pipeline
The data analytics pipeline is the series of phases that moves raw data from its sources to the people who use the results. Each phase's output is the next phase's input.
Iterations and interactions: analysis can show that data is missing or poor quality, sending the work back to capture or processing; feedback from users of a dashboard can change what is captured; new data keeps flowing in, so the pipeline runs continuously rather than once.
DLM describes the life of the data itself (creation to destruction). The pipeline describes how data is turned into insight (capture to use). Questions on job roles (Topic 5) expect you to say which phase each role works in.
2.2.1 Data assurance considerations
Data assurance means making sure data can be trusted. Each consideration below increases confidence in the data; failing on any one affects how it can be collected and used.
Validation checks data is sensible; verification checks it matches the source. Neither guarantees accuracy on its own, which is why both are used.
2.2.2 Data gathering
Data gathering is the process of collecting data from sources to answer a question or meet a requirement. The process: define the purpose and the data needed, choose a method and a sample, collect the data, check its quality (validate, verify, clean), then store it securely.
Meeting its purpose and good quality: check the data answers the requirement, comes from a reliable source, is accurate, complete, up to date and unbiased, and has been validated and verified.
Consequences of poor-quality data: wrong conclusions and poor decisions, wasted money and time, lost customers and reputation, and possible legal problems (inaccurate personal data breaks the accuracy principle).
For "explain two benefits of gathering data this way", use the scenario every time. OCR's model answer for street interviews: "many different people pass through the day (1) so the researcher gains a variety of opinions (1)". Generic answers are rejected.
2.3.1 Data states
One piece of data passes through all three states: a customer record is at rest in the database, in transit when sent to a sales assistant's device, and in use while it is open on screen.
2.3.2 Data stores
A data store is a repository for persistently storing and managing collections of data, such as a database, a file system, a spreadsheet, a data warehouse or a cloud storage service.
Purpose: keep data safely so it is not lost when power is off; organise it so it can be found, searched and retrieved; control who can access it; and make it available to the applications and people that need it.
How data stores interact: data is passed between stores, for example a sales database feeds a data warehouse each night, a spreadsheet imports a CSV export from a database, or an application reads from a cloud store through an API. In the pipeline, the capture phase writes into stores and the analysis and use phases read from them. Interaction needs compatible formats and secure connections.
2.3.3 Data storage types: block, file and object
Factors when selecting: speed needed, how often data changes, volume and growth, cost, how it will be accessed and shared, and security requirements.
2.3.4 On-site storage
Factors when selecting on-site storage: capacity, speed, cost, number of users, need for remote access, scalability, security, and whether the organisation has IT staff.
OCR's 6-mark "Should the company use a NAS?" question is levels-marked. Top band needs benefits and limitations applied to the scenario, plus a clear conclusion with reasons. For a seven-person office with no IT staff, credited points include: simple to set up, supports office applications, easy collaboration, cheaper than a large server, limited scalability if staff grow, and needs physical security.
2.3.5 Cloud storage
Cloud-based database services (Database as a Service, DBaaS): a provider hosts and manages a database in the cloud, handling installation, backups, updates and scaling. Uses: website and app back ends, storing customer and transaction data, analytics on large data sets, and sharing data between sites. The organisation pays for what it uses and does not need database servers of its own.
Factors when selecting: sensitivity of the data, legal requirements (where data is stored), cost, scalability, reliability and uptime, security offered, internet connectivity, and the skills of the organisation's staff.
OCR's indicative content for "analyse the impacts of a public cloud" on a new company includes: no hardware investment; flexible plans; large providers employ experienced security staff; risk of downtime and denial-of-service; loss of control over data; reliance on the provider for backups; multitenancy increasing latency; and data stored abroad raising legal issues.
2.4 Data transformation: wrangling and maintenance
Importance: raw data is often messy, inconsistent and in different formats. Wrangling makes it accurate and consistent, so analysis gives reliable results. It is time-consuming but essential.
Importance: data goes out of date (people move, prices change). Maintenance keeps decisions based on current data, avoids errors and duplicates building up, meets the legal duty to keep personal data accurate, and keeps data useful.
Not in F200: specific data wrangling activities and specific data maintenance activities. (F201 covers data cleaning techniques.)
2.5 Usage and analysis: data analytics
Data analytics is examining data to draw conclusions from it. Its purpose is to apply statistical analysis and technologies to data to find trends and solve problems, so organisations can make informed decisions. (These two words complete the sentence on OCR's sample paper.)
The four main types build on each other: descriptive (what), diagnostic (why), predictive (what next), prescriptive (what to do). Cognitive adds AI on top. Match the type to the question the scenario is asking.
2.6.1 Presenting data
Criteria for selecting a presentation method: the audience and their expertise, the purpose (compare, show a trend, show exact values, explain), the type and amount of data, accessibility (colour blindness, screen readers), and how it will be delivered (printed, on screen, in a meeting).
2.6.2 Visualising data
Data visualisation communicates information by turning data into visual forms, so the audience can understand trends, patterns and comparisons quickly.
Criteria for selecting a visualisation method: what the data shows (comparison, trend, proportion, relationship), the audience, the amount of data, whether it must update live, the medium, and accessibility.
Sample paper: a college wants to show what proportion of students travel by car, bus or train. Answer: pie chart, because it shows proportions of the total and suits a small data set.
2.7 Archival
Archiving moves data that is no longer in active use to separate long-term storage, where it is kept safely and can be retrieved if needed.
Importance: legal and regulatory requirements to keep records (financial, medical, employment records); evidence for audits and disputes; historical analysis; frees fast, expensive primary storage and speeds up live systems; reduces the amount of data exposed in day-to-day use.
How data is archived: identify data past its active period using a retention policy, copy it to archive storage with metadata so it can be found, check the copy, remove it from the live system, and protect it with access controls and encryption.
2.8 Destruction
Importance of data destruction: UK GDPR's storage limitation principle says personal data must not be kept longer than needed; deleting a file normally only removes the pointer, so data can be recovered from discarded drives; secure destruction prevents breaches, identity theft and fines; and it frees storage.
Not in F200: non-secure destruction methods (such as simply deleting files or emptying the recycle bin).
Match the method to the media and whether the drive will be reused. Degaussing is useless on an SSD; erasure or overwriting keeps the drive; destruction and shredding are for drives that will never be reused.
Topic 3 - How data can be accessed and managed across platforms
3.1 Application Programming Interfaces (API)
An API is a set of rules that lets one application communicate with another, defining how requests for data or services are made and how responses are returned. Role: APIs let data be shared and accessed across platforms (web, mobile, cloud) without each system knowing how the other works inside, and they control what data is exposed, which supports security.
API certifications (who is allowed to use the API):
API types and the data format each uses:
Not in F200: a detailed understanding of how the different API types work.
Sample paper match: "provides restricted access, intended for specific purposes" is Partner; "allows external users to access data or services" is Public.
3.2 User access controls
Sample paper: a government office storing highly sensitive data should use MAC. OCR credits reasons such as "access based on the user's level of clearance", "MAC is the most secure model", "access granted on a need-to-know basis" and "access cannot be altered by any user".
3.3 Permissions
A user works with data within the rights they have been given. An administrator controls the system: they allocate access controls and privileges to users, and so decide what each user can do.
Administrator rights:
- Allocating access controls: choosing and applying the access control model (MAC, DAC, RBAC, ABAC, RuBAC) and setting who can reach which systems.
- Allocating user privileges at three levels:
- User level: rights given to one individual account, e.g. one manager can approve refunds.
- User group level: rights given to a group, and every member inherits them, e.g. all Sales staff can read the customer database. Quicker to manage for many users.
- File and folder level: rights attached to a specific file or folder, e.g. only HR can open the Payroll folder. Gives the finest control over sensitive data.
Sample paper: a database administrator who adds data and pictures to a stock file needs rights such as read (to access the stock file) and write (to create new entries). Each right needs its own reason linked to the duty.
Topic 4 - Legal considerations
4.1 Legislation and the role of the ICO
Main purpose: makes unauthorised access to computer systems and data a criminal offence, so it protects data held by organisations from hackers.
Important aspects: offences of unauthorised access to computer material; unauthorised access with intent to commit a further crime; unauthorised acts that impair a computer (including spreading malware and deleting data); and making or supplying tools for these offences.
Complying: only access data and systems you are authorised to; set and log access rights; get written permission before any testing.
Non-compliance: criminal prosecution, fines and imprisonment.
Main purpose: controls how personal data is used by organisations, businesses and government, and gives individuals rights over their data. It is the UK's implementation of data protection law alongside the UK GDPR.
Complying: only collect personal data with a lawful reason; keep it accurate, secure and no longer than needed; answer subject access requests; report serious breaches to the ICO.
Non-compliance: ICO enforcement notices and fines, compensation claims, reputational damage.
Main purpose: gives people a right of access to recorded information held by public authorities (government departments, councils, NHS, state schools, police).
Important aspects: anyone can make a written request; the authority must normally respond within 20 working days; some information is exempt (e.g. personal data, national security); public authorities must publish certain information proactively.
Complying: keep records organised so they can be found, and respond on time.
Non-compliance: complaints to the ICO, decision and enforcement notices; destroying requested information to prevent disclosure is a criminal offence.
Main purpose: protects individuals' personal data by setting the key principles, rights and obligations for processing it.
Important aspects: seven principles (lawfulness, fairness and transparency; purpose limitation; data minimisation; accuracy; storage limitation; integrity and confidentiality; accountability); rights of individuals such as access and erasure; breaches reported to the ICO within 72 hours.
Non-compliance: fines up to £17.5 million or 4% of annual global turnover, whichever is higher.
Main purpose: gives people specific privacy rights in relation to electronic communications.
Important aspects: rules on marketing by email, text and phone call (consent is usually needed before sending marketing emails or texts to individuals); cookies and similar tracking need consent unless strictly necessary; security of public communications services; rules on traffic and location data.
Complying: get opt-in consent for electronic marketing and non-essential cookies, offer an easy opt-out in every message, check numbers against the Telephone Preference Service.
Non-compliance: ICO fines and enforcement notices.
Role: gives guidance to organisations and the public; keeps a register of organisations that pay the data protection fee; handles complaints; investigates breaches; carries out audits; issues warnings, reprimands, enforcement notices and fines; and can prosecute criminal offences.
Not in F200: the detailed content of each act or regulation. You need the latest version, the main purpose and important aspects, how to comply, the impact of non-compliance and the role of the ICO.
Sample paper match: privacy rights in electronic communications = PECR; protects data held by companies from hackers = Computer Misuse Act; right of access to information held by UK public authorities = Freedom of Information Act.
Topic 5 - Job roles, skills and attributes in data analytics
5.1 Job roles related to data analytics
Not in F200: the specific skills required for each job role, and detailed lists of responsibilities.
The two definitions in bold above are the exact phrases OCR's sample mark scheme uses for Data Analyst and Data Engineer. Do not confuse them: the analyst studies the data; the engineer builds the systems that move and transform it.
5.2 Personal attributes
The sample paper asks for one attribute for an AI Scientist and one reason. OCR credits problem solving (to design models and algorithms), team working (to work with other staff) and effective communicator (to communicate findings to others). The reason must match the attribute you gave.
5.3 Communication skills
Sample paper: a database administrator's communication skill. Credited answers: verbal (for meetings with colleagues), written (for entering data correctly) or questioning techniques (to elicit specific information).
Topic 1 - The scope of managing big data
1.1 The six characteristics (6Vs)
The 6Vs are the characteristics that define data as big data. Use the spec's full wording for each one: OCR's sample mark scheme accepts "Variety of data types" but rejects "Variety" on its own.
Sample paper: "state two ways the volume of data can characterise data as big data". OCR credits: the amount of data that exists; the initial size of data collected; the amount of data is growing.
1.1 Six steps for analysing big data
Learn the order. Cleaning comes before mining: mining dirty data finds false patterns.
1.2 The evolution of big data
1.3 How big data is captured
Big data capture is collecting data from its sources and bringing it into a system so it can be stored and analysed.
1.4 The purpose, importance and use of big data analytics
F201 has one scenario that develops through the paper (the sample uses a large retailer with a loyalty scheme). Every PO2 answer needs a second clause applying the point to that scenario, e.g. "stock levels can be monitored (1) so the retailer can order more of its popular products (1)".
Topic 2 - The infrastructure challenges of big data
2.1 Types of big data
Sample paper: "transactional data is structured data; state two other examples". Credited: continuous data, discrete data, relational database, spreadsheet file data. Repeating transactional data gets no mark.
2.2 Data preparation and cleaning for data mining
Purpose of preparing data: to get raw data into a consistent, accurate form so mining and analysis give reliable results.
- Numeric data is prepared by setting consistent units and decimal places, converting numbers stored as text into numbers, dealing with outliers and missing values, and scaling values so they can be compared.
- Textual data is prepared by making case and spelling consistent, removing extra spaces and punctuation, splitting text into words or fields, translating, and grouping words with the same meaning.
Data wrangling transforms and prepares raw data from one or more sources into a format that can be analysed or used by a computer system or application. It can be time-consuming but is essential for obtaining reliable results from data analysis. It is used to prepare data by combining sources, restructuring, cleaning, enriching and validating it. (The three bold words complete the sample paper's sentence.)
Purpose of cleaning data: to remove or correct inaccurate, incomplete, duplicated and irrelevant data so the results can be trusted.
Not in F201: technical details of each data cleaning technique.
Sample paper: loyalty records with one customer listed twice, a missing postcode and an email address without an @. Credited techniques: removing duplicates, fix missing data, fix structural errors. Read the data given and match each problem to a technique.
2.3 Data mining techniques
Data mining is the process of searching large data sets to discover hidden patterns, relationships and anomalies that can be used to make decisions. Its role in analysing big data is to find the patterns that would be impossible to spot by hand.
Not in F201: coding details.
2.4 Big data infrastructure: server configurations
When used: moderate data volumes, sensitive data, predictable workloads.
Benefits: full control, consistent performance, good security because nothing is shared.
Limitations: limited capacity; scaling means buying a bigger machine; single point of failure.
When used: very large volumes and high velocity data, e.g. search engines, streaming services.
Benefits: scales by adding nodes; processes in parallel so is faster; fault tolerant because work moves to other nodes if one fails.
Limitations: complex to set up and manage; needs specialist staff; network delays between nodes.
Factors influencing the choice: volume and growth of data, velocity (real-time needs), processing power required, reliability and fault tolerance, security and compliance needs, scalability, staff skills available.
Not in F201: costs of implementation.
2.4 Big data infrastructure: software platforms
OCR's sample mark scheme for "one benefit and one limitation of data visualisation software" credits, for example: charts and graphs can be created so the retailer can visualise data rather than look at raw data; and the output may be cluttered or not fit for the audience, so the retailer struggles to extract the relevant results.
2.4 Big data infrastructure: data storage areas
Not in F201: providers of data lakes and data warehouses.
Lake = raw data in any format, stored first and shaped later. Warehouse = cleaned, structured data, shaped first. Sample paper credits any of public cloud, data lake, data warehouse or SSDs as "another storage area" to private cloud.
2.5 Data science and data analytics
Difference: data science builds the models and tools and looks for questions worth asking; data analytics applies tools to existing data to answer a defined question. Data analytics is often treated as part of data science.
Use in the areas of application (1.4): in healthcare, data science builds diagnostic models from scans while analytics reports waiting times; in retail, data science builds the recommendation engine while analytics finds the best-selling products by region; in banking, data science builds fraud models while analytics reports monthly fraud rates.
2.6 Data analytic techniques
Not in F201: the technical details of each technique.
The sample paper's 9-mark question was "discuss whether the retailer should use cluster analysis". Level 3 (7 to 9 marks) needs a range of examples of its use, benefits and limitations for the retailer, analysis in context and a clear conclusion with reasons. OCR's examples: links between products bought and age range, location, and online versus in store.
Topic 3 - Big data, machine learning and artificial intelligence
3.1 Artificial intelligence (AI)
Artificial intelligence (AI) is the ability of computer systems to perform tasks that normally need human intelligence, such as recognising patterns, understanding language, making decisions and learning from experience.
How AI is developed and used in data science: developers collect and prepare large training data sets, choose an algorithm, train a model on the data, test it on new data, then deploy and monitor it. In data science AI is used to find patterns in huge data sets, analyse unstructured data (images, speech, text), make predictions and recommendations, and automate decisions.
Sample paper: "explain how AI can be used to provide recommendations". Credited: algorithms consider the customer's transaction history (1) and select products based on what was previously bought (1).
3.1 Machine learning and its algorithms
Machine learning (ML) is a branch of AI in which systems learn from data to identify patterns and make predictions or decisions, improving with experience without being explicitly programmed for each task. ML algorithms are trained on historical data, then applied to new data.
How big data, ML and AI interact in data science: big data provides the large, varied training data that machine learning needs; ML is the method by which AI systems learn from that data; AI then applies what has been learned to analyse new data and make predictions or decisions, which creates more data that can be used to retrain the model. Without big data, models are inaccurate; without ML and AI, big data is too large to analyse by hand.
Not in F201: the historical development of AI and machine learning.
Random Forest is many decision trees combined; K-Means is clustering. Decision Tree and Random Forest predict a known type of answer from labelled examples; K-Means finds groups that nobody labelled.
Topic 4 - Legal and ethical issues in data management
4.1 Legal issues: UK GDPR
Latest version: the UK GDPR is the retained EU GDPR as it applies in UK law from 1 January 2021, alongside the Data Protection Act 2018, as amended (including by the Data (Use and Access) Act 2025).
Main purpose: to protect an individual's personal data, and to set out the key principles, rights and obligations for processing personal data. (Both wordings are credited on the sample mark scheme.)
How to comply: identify a lawful basis; collect only what is needed; keep data accurate; set retention periods; secure data with encryption and access controls; train staff; keep records of processing; carry out a Data Protection Impact Assessment (DPIA) for high-risk processing; respond to rights requests; report breaches within 72 hours.
Why compliance is important: it is a legal requirement; it protects individuals from harm; it builds customer trust; and it avoids penalties.
Impacts of non-compliance on organisations: fines of up to £17.5 million or 4% of annual global turnover, whichever is higher; increased costs for staff training and security; reputational damage and lost customer trust; legal action and compensation claims; business disruption; regulatory investigations by the ICO.
How organisational policies mitigate non-compliance: a data protection policy, retention and destruction policy, access control and acceptable use policies, breach response procedure, consent and marketing policy, and regular training and audits make compliance a routine part of work and provide evidence for accountability.
Not in F201: the detailed content of the regulation.
OCR's sample mark scheme quoted the EU figure of 20 million (euros). The UK GDPR maximum is £17.5 million or 4% of annual global turnover, whichever is higher. Use the UK figure, and in a 6-mark analysis tie each impact to the scenario (e.g. 250 stores means many staff to retrain).
4.2 Ethical issues: automated decision making
Automated decision making is making a decision by technological means (an algorithm) without human involvement, e.g. approving a loan, shortlisting job applicants, choosing who receives an offer.
How it creates risks of discrimination and bias:
- A lack of training data for certain groups means the algorithm has not learned their patterns, so it treats them poorly.
- The person creating the algorithm has unconscious bias, which is built into the final algorithm.
- Historical data used for training already contains past bias, which the algorithm learns and repeats.
How UK GDPR applies in organisations: people must be told when significant decisions about them are made solely by automated means, including profiling. Where such a decision has legal or similarly significant effects, the organisation must provide safeguards: the person can make representations, obtain human intervention and contest the decision. Special category data has tighter restrictions. A DPIA is expected for this kind of high-risk processing.
Impacts on individuals: unfair refusal of credit, jobs, insurance or services; discrimination against protected groups; not knowing why a decision was made; difficulty challenging it; on the positive side, faster and more consistent decisions.
The sample paper's 2-mark "explain how automated decision making could increase the risk of bias" needs a cause and an implication, e.g. "historical data used to train the algorithm may include bias (1), which is built into the patterns the algorithm learns (1)".
4.2 Ethical issues: collection, storage, ownership and sharing
Protecting the identity of individuals when collecting, storing and sharing data: collect only what is needed; anonymise data (remove identifying details so people cannot be identified) or pseudonymise it (replace identifiers with codes, kept separately); aggregate results so individuals are not visible; encrypt and access-control stored data; share only under agreements and only what the recipient needs.
How data ownership is determined: by who created or collected the data, contracts and terms and conditions (e.g. a platform's terms on user content), intellectual property and copyright law, and employment contracts (work created for an employer usually belongs to the employer). Individuals do not "own" personal data held about them in a legal sense, but UK GDPR gives them rights over it, and the organisation is the controller responsible for it.
How organisations respond: privacy notices explaining what is collected and why; consent management; data sharing agreements; retention and destruction schedules; appointing a Data Protection Officer; DPIAs before new projects; ethics review boards; staff training.
4.2 Emerging ethical debates and frameworks for ethical data management
Emerging ethical debates affecting the use of data include: facial recognition in public places; AI trained on people's data or creative work without permission; the extent of workplace monitoring; selling or sharing location and health data; children's data on apps and social media; generative AI producing false or harmful content; and who is accountable when an algorithm makes a harmful decision.
Impacts on individuals: loss of privacy, feeling watched, discrimination, misuse of their data or work, and loss of trust in organisations. How organisations respond: publish AI and data ethics policies, carry out impact assessments, keep a human involved in significant decisions, be transparent about how algorithms are used, and allow people to opt out.
Sample paper: a retailer selects customers for a questionnaire using transaction records. "All population groups must be included" is needed so views are representative and do not show bias against a group; "up to date" so questionnaires go to current members who actually bought the product.
Topic 5 - Environment and society
5.1 Environment
Sample paper: one benefit and one limitation of big data for warehouse energy efficiency. Credited benefit: analysis shows when and where the warehouse is used (1) so energy can be managed in those areas at those times (1). Credited limitation: patterns from normal working days may change (1), so the algorithm is wrong and energy is wasted (1).
5.2 Society
A smart city is an urban area that uses data from connected sensors, devices and systems to manage resources and services efficiently and improve the quality of life of the people living there.
How data from a variety of sources is exchanged to optimise city operations: traffic sensors, CCTV, public transport GPS, smart meters, waste bin sensors, air quality monitors and residents' apps send data over networks to central platforms. The data is shared between services (transport, energy, police, councils), analysed in real time, and used to adjust systems automatically, e.g. changing traffic light timings, rerouting buses or scheduling bin collections only when bins are full.