This programme is not about starting from zero. Every one of you brings real-world experience — industry knowledge, problem-solving habits, and an understanding of how organisations work. Data science is a set of structured thinking tools that lets you apply that experience more powerfully.

Over five days you will learn to ask sharper questions, work confidently with evidence, prepare and evaluate information, use AI tools responsibly, and communicate findings clearly. These are practical employability capabilities that employers across every sector value.

Data science is not just programming or statistics. At its core, it is a disciplined way of turning raw, messy information into decisions that are grounded in evidence rather than gut feeling alone. It combines curiosity, critical thinking, domain knowledge and — yes — some tools and techniques.

Look at the three statements below, drawn from the fictional JobSeeker360 Dataset. Discuss in pairs and decide how to classify each one.

The first instinct is to ask "How do we help people get jobs?" That is too broad to answer with data. Good data science begins with a precise, measurable and responsible question.

You have a budget of ten tokens. Bid on the questions you believe are most useful and answerable with the JobSeeker360 Dataset. Discuss why some questions are more precise, more responsible or more actionable than others.

Each participant receives one card printed with a lifecycle stage. Working together, arrange the eight stages in the correct sequence and agree on what each stage produces as an output. Two mistake cards are hidden in the deck:

Before you can analyse data, you need to understand what kind of data you are working with. Different types require different handling and different analytical techniques.

Before closing this session, each participant drafts a personal problem statement using the template below. This will reopen the next session and anchor all the data-preparation work to a real question.

We begin Day 2 by returning to the problem statement you drafted yesterday. Before any data can be collected, cleaned or analysed, the question must be clear enough to define exactly what data you need and what quality that data must meet.

This is a synthetic, fictional dataset created for training purposes only. It does not represent actual real participant records, real participants or real outcomes. It is designed to look realistic so that you can practise genuine data-preparation skills.

You receive a copy of the JobSeeker360 Dataset that has not been validated. It contains six categories of problem, each of which could distort any analysis built on top of it.

Working in three pairs, each team investigates a different category of data problem in the synthetic spreadsheet. You have 20 minutes to find as many issues as possible, document what you find and propose a resolution.

Not every field in a dataset earns its place. Some fields are essential to answering the question; some are useful but not critical; some are sensitive and require justification; and some add noise without adding insight. Review each field in the JobSeeker360 Dataset and assign it a status.

A data dictionary is the contract between the people who create the data and the people who analyse it. Without it, two analysts can interpret the same field in entirely different ways and reach entirely different conclusions.

Five fields from the JobSeeker360 Dataset go on trial. Each round, four participants take a role and debate the correct definition before the group reaches a verdict.

Upload the synthetic JobSeeker360 Dataset to ChatGPT Data Analysis. Ask it to report on data quality before any analysis begins. A responsible inspection prompt looks like this:

ChatGPT should return a structured report. Review it carefully: it may miss issues, invent plausible-sounding values or misidentify data types. Treat its output as a first draft, not a final audit.

A Custom GPT is a version of ChatGPT configured with specific instructions so that it behaves consistently for a defined task. The fictional Data Readiness Coach is configured with four standing rules:

Gemini Notebook is a research and briefing workspace. You upload reference documents and ask questions that Gemini answers using only those sources — citing each claim back to its origin. This makes it ideal for learning, summarising and preparing briefing notes from trusted materials.

For this programme, upload the DS2010 course outline, the synthetic data dictionary, a short methodology note and a permitted reference document. Gemini Notebook is not the primary numerical data-analysis engine — use ChatGPT or Manus for calculations and statistics.

Manus is an AI agent that can handle an end-to-end workflow: inspect data, identify patterns, generate visualisations, write explanatory text and produce a presentation-ready brief — all from a single structured prompt.

Questions 5–8

At the end of Day 2, each group prepares a data-readiness note — a short, structured document that certifies the state of the dataset before it is handed to the next trainer. This mirrors professional data-handover practice.

Working as a full group, complete the data-readiness note for the synthetic JobSeeker360 Dataset. Each pair contributes one section based on their Escape Room findings. The group then reads the note aloud and votes: Ready, Ready with conditions, or Not ready — more work needed.

Exploratory Data Analysis (EDA) is the stage where you look systematically at patterns, distributions, relationships and anomalies in the cleaned dataset. EDA asks what is happening — it does not automatically tell you why it is happening. Interpretation requires domain knowledge and careful reasoning.

In Day 4, you will learn the principles of effective data visualisation: choosing the right chart type, designing for clarity, and avoiding common misrepresentations. As a preview, consider how one validated finding from EDA might be communicated visually.

Suppose EDA reveals that participants who used online support channels were placed faster than those who used in-person channels. A simple grouped bar chart — placement status by support channel — communicates this finding in seconds to a decision-maker who has not read the analysis. Using a tool such as Manus, you can generate a draft of this chart from your prepared data and then refine it in Day 4.

Machine learning (ML) uses patterns in historical data to make predictions about new cases. In our career transition context, a simple classification model might prioritise which participants may benefit most from additional support — based on features such as training completion, number of applications and support channel used.

Copy, adapt and reuse these prompts in your own projects. Always replace the bracketed placeholders with your actual context.

Copy, adapt and reuse these prompts in your own projects. Always replace the bracketed placeholders with your actual context.

Take ten minutes before leaving today to reflect on your learning journey so far. Use the questions below as prompts — honest reflection is one of the most effective learning strategies you can practise.

You have built a solid foundation over the sessions so far. Here is what the appointed specialist trainers will take you through next, building directly on the cleaned dataset and problem statement you have prepared.

In data science, diverse industry backgrounds are a profound strength. Your unique experiences provide invaluable perspectives, helping you identify critical insights and ask pertinent questions, enhancing outcomes beyond what a pure technician might achieve.

Gemini Notebook (formerly NotebookLM) lets you upload multiple documents and ask questions that are answered only from those sources — with citations. It now also runs code for data analysis (Ultra/Pro users).

Each tool has a sweet spot. Use this as a quick reference when you are deciding which AI to reach for.