B1.1 Approaches to Computational Thinking
Revision notes. Everything in the syllabus, nothing extra.
Contents
Computational thinking is a way of approaching a problem so that its solution can be carried out by a computer. It rests on four concepts:
| Concept | In one line |
|---|---|
| Decomposition | Break a big problem into smaller parts. |
| Pattern recognition | Spot similarities so one solution can be reused. |
| Abstraction | Remove unnecessary detail, keep what matters. |
| Algorithmic design | Write clear step by step instructions to solve it. |
Break it, spot it, simplify it, solve it.
B1.1.1: Problem Specification
A problem specification defines exactly what must be solved before anyone starts solving it. It has six parts. Every part must be clear, specific and measurable.
| Part | What it says | Weak | Strong |
|---|---|---|---|
| 1. Problem statement | What the problem is and why it matters. Concise, specific, no ambiguity. | "Our customer service is inefficient." | "Customers wait an average of 25 minutes on hold, causing a 15% rise in abandoned calls last quarter." |
| 2. Constraints and limitations | Restrictions on the solution: resources (time, budget, staff), technical (hardware, software), external (laws, other systems). | "We don't have many resources." | "The budget is limited to $10,000 and cannot be exceeded due to funding rules." |
| 3. Objectives and goals | Objective: broad, high level aim. Goals: specific, measurable, time limited targets that achieve it. | "Make customers happier." | Objective: reduce wait times. Goal: cut average hold time by 20% within the next quarter. |
| 4. Input specification | The format, type and expected characteristics of the data going in. | "Customer feedback." | Star ratings 1 to 5, free text comments and NPS scores, collected straight after the interaction and linked to the customer. |
| 5. Output specification | The format, content and presentation of the results coming out. | "Some reports." | A dashboard with weekly satisfaction charts, filterable by date and channel, plus alerts for very negative feedback. |
| 6. Evaluation criteria | The benchmarks used to judge whether the solution worked. | "It works well." | The chatbot resolves 70% of enquiries without a human and scores 4/5 on post chat surveys. |
NPS (net promoter score) rates how likely a customer is to recommend the company to a friend.
Evaluation criteria: the five factors
| Factor | Question | Chatbot example |
|---|---|---|
| Effectiveness | Does it solve the problem? | % of enquiries resolved without a human |
| Efficiency | Does it save time or resources? | Faster resolution, available 24/7, less load on staff |
| Accuracy | Are the results correct? | Correct answers, understands different phrasings of the same request |
| Usability | Is it easy to use? | Simple interface, clear prompts, little training needed |
| Maintainability | Can it be updated and adapted? | New questions or channels can be added easily, data kept safe |
Ninja Notes:
- Learn the six parts in order: Statement, Constraints, Objectives and goals, Inputs, Outputs, Evaluation.
- If it can't be measured, it's not a good goal. Numbers plus a time limit.
- Goals become your evaluation criteria at the end. This is exactly how your IA works.
B1.1.2: The Four Concepts
Decomposition
Definition: breaking a complex problem down into smaller, more manageable parts that can each be understood and solved separately.
- Single responsibility principle (SRP): each part should do one clear job.
- Ask: Can it be split into steps or phases? Are there repeated parts to group? Which parts are independent? What does a diagram of it look like? Could I explain it to a five year old (ELI5)?
- Examples: a bicycle into frame, drivetrain, wheels, steering, brakes, seat, gears, suspension, lights, fenders. A smartphone into screen, battery, charging circuit, CPU, RAM, storage, buttons, ports, cameras, sensors, OS.
| Decomposing a game NPC (non playing character) | |
|---|---|
| Sensing | Perceive the player and surroundings (sight, hearing). |
| Decision making | Evaluate the situation and choose an action (behaviour tree, finite state machine). |
| Actions | Patrol, attack, flee, hide. |
| Pathfinding | Navigate the world to reach a goal. |
The textbook also shows a deeper split: motivation (goals, personality), perception (sensory input, filtering), decision making (behaviour library, evaluation module), action execution (animation, movement). Both are valid. Decomposition keeps going until each part is simple.
Pattern recognition
Definition: identifying similarities, trends or regularities within or between problems, so that a solution to one can be generalised and reused instead of solving each variation from scratch.
- Ask: What is repeated? How do things change over time? Do different parts do the same job? Could specific values be replaced with variables?
- Example 1: 2, 4, 6, 8. Each number is 2 more than the last, so
next = current + 2. One rule predicts any term. - Example 2: handwritten digits. An image is a grid of pixel brightness values. A "1" is a vertical line, an "8" is two loops, so any shape with a loop is likely 0, 6, 8 or 9.
Abstraction
Definition: filtering out unnecessary details to focus on the essential features of a problem, and representing it as a simplified model.
- The hard part: deciding what is essential. A car's engine matters, its colour doesn't.
- Ask: What is the core purpose? What are the minimum inputs and outputs? If I remove this part, does it still work? Can several steps become one higher level idea?
| Predicting a winner | Keep | Ignore |
|---|---|---|
| Football match | Serious injuries, past results between the teams | Shoelace colour, probably player age |
| Car race | Car performance (engine, tyres, aerodynamics), driver skill, track suitability, weather, chance of mechanical failure | Car colour, driver's favourite snack, cloud shapes |
The race model becomes a few scores out of 10 plus failure chances. It is a simplification, but a useful one. That is the essence of abstraction.
Algorithmic design
Definition: developing a step by step set of instructions (an algorithm) that solves a problem. It must be clear, precise, unambiguous and efficient.
- Ask: What exactly is the problem? What are the inputs? What are the outputs? Where are the decisions? Where are the loops? What is the order of steps?
- Can be written as code, pseudocode or a flowchart.
- Test: explain it step by step to a friend. If they get stuck, a step is missing.
B1.1.3: Applying Computational Thinking
In an "explain how" question, go through all four concepts and give one specific example of each in the scenario.
| Field | Decomposition | Pattern recognition | Abstraction | Algorithmic design |
|---|---|---|---|---|
| Software development (CRM system) | Modules: lead tracking, customer management, communication | Recording common bugs to speed up future debugging | OOP objects with reusable methods | Exact rules and steps for automatic email scheduling |
| Data analysis (sales data) | Cleaning, normalisation, analysis | Seasonal sales trends to predict demand | Focus on KPIs, filter out irrelevant data | ETL pipeline: extract, transform, load |
| Machine learning | Data collection, features, training, evaluation, deployment | Finding image features that classify objects (CNNs) | Data represented as features and labels | Network design: layers, activation functions, optimiser |
| Database design | Conceptual, logical, physical stages | Common queries, so index or denormalise for them | Tables that model real entities, ignore irrelevant detail | Efficient querying, updating, transactions for integrity |
| Network security | Physical, infrastructure, application security | Known attack patterns (DDoS, phishing) | Generic security models and protocols | Encryption algorithms and protocols (TLS) |
B1.1.4: Tracing Flowcharts
A flowchart is a diagram that shows the steps, decisions and flow of control of an algorithm using standard symbols (ISO 5807). It is read top to bottom, left to right.
- Decision: exactly two exits, labelled Yes/No or True/False. The question inside must have only those two answers.
- Process: changes the value, form or location of data.
- Connector: joins parts of a big flowchart without drawing a long line.
How to trace
- Draw a trace table: one column per variable, one for each decision, one for output.
- Follow the arrows exactly. Write a new row every time a value changes.
- At every diamond, work out the condition with the current values and follow that exit.
- Stop at End. Check the last pass, where the decision finally flips. Off by one mistakes live there.
Example 1: a simple decision
Example 2: a loop (guess the number)
The "No" path loops back to the input, so the player keeps guessing until the decision becomes "Yes".
Example 3: linear search
Checks every element in turn from index 0 until it finds the target or reaches the end. Array: 2, 4, 6, ... 100. Target: 42.
42 sits at index 20, so the loop checks indexes 0 to 20 and outputs "Found at 20" after 21 comparisons. Searching for 43 would check all 50 elements, then output "Not found".
Example 4: how many iterations?
| Iteration | A | B | A > B? |
|---|---|---|---|
| start | 10 | 100 | No |
| 1 | 20 | 90 | No |
| 2 | 30 | 80 | No |
| 3 | 40 | 70 | No |
| 4 | 50 | 60 | No |
| 5 | 60 | 50 | Yes → End |
5 iterations. The loop body runs 5 times, but the decision is checked 6 times. Read the question carefully to see which one it asks for.
Practice
- Splitting a game into a menu screen, a play screen and a score screen.
- Ignoring a car's colour when predicting a race winner.
- Noticing three parts of a program all sort a list, so writing one sort function.
- Writing the exact steps for when an email reminder is sent.
a. Decomposition b. Abstraction c. Pattern recognition d. Algorithmic design
"The school website is bad and students do not like it."
"Students need an average of 5 clicks to find the timetable on the school website, and 60% of surveyed students could not find it at all, which has doubled timetable emails to the office." Specific problem, numbers, and why it matters.
A school canteen has long queues at lunch.
Objective: reduce the time students spend queuing for lunch.
Goals: cut the average queue time from 15 to 5 minutes by the end of term. Serve 90% of students within the first 20 minutes of lunch by January. Both are measurable and time limited.
Goals: cut the average queue time from 15 to 5 minutes by the end of term. Serve 90% of students within the first 20 minutes of lunch by January. Both are measurable and time limited.
Removing / filtering out unnecessary details [1] so that only the essential features of a problem remain, giving a simplified model [1].
Set x = 12 and y = 8. While x ≠ y: if x > y then x = x − y, otherwise y = y − x. When they are equal, output x.
| x | y | x ≠ y? | Action |
|---|---|---|---|
| 12 | 8 | Yes | x = 12 − 8 = 4 |
| 4 | 8 | Yes | y = 8 − 4 = 4 |
| 4 | 4 | No | Output 4 |
Set total = 0 and n = 1. While total < 20: total = total + n, then n = n + 2. Output total and n.
| Iteration | total | n | total < 20? |
|---|---|---|---|
| start | 0 | 1 | Yes |
| 1 | 1 | 3 | Yes |
| 2 | 4 | 5 | Yes |
| 3 | 9 | 7 | Yes |
| 4 | 16 | 9 | Yes |
| 5 | 25 | 11 | No |
Ninja Notes:
- Four concepts: decomposition, pattern recognition, abstraction, algorithmic design. Break it, spot it, simplify it, solve it.
- Six spec parts: statement, constraints, objectives and goals, inputs, outputs, evaluation criteria.
- Objective = broad aim. Goal = measurable target with a deadline.
- Evaluation criteria: effectiveness, efficiency, accuracy, usability, maintainability.
- "Explain how computational thinking is applied" means all four concepts, each with an example from the scenario.
- Decision diamonds have exactly two exits. Label them.
- When tracing, use a table and check the last pass of every loop.