You are viewing a public preview
Sign in to save your progress, do assignments, join discussions and get feedback.
Identifying problems and analytical questions; determining information needs; collecting and organizing data; selecting visual methods; building, checking, refining, interpreting and communicating visual outputs.
Learning outcomes
The lecture slides for this lesson are below.
Review the data concepts and data types; read about visualization principles, chart selection and the visualization process; for a few business questions from your own work or study, decide which chart would answer each one and explain why. (About 12 hours of independent study.)
By the end of this lesson you can:
Key idea: Visualization is a process, not a single step. You define requirements, prepare data, analyse and visualise, then iterate and refine until the chart reliably answers the business question.
The lecture presents the data visualization process as four stages arranged in a circle, with an arrow labelled "Iterate & Refine" linking the end back to the start:
| Stage | Focus | Typical output |
|---|---|---|
| 1. Define requirements | Objectives and scope | Business problem, analytical questions, audience, metrics |
| 2. Collect and prepare data | Data sources and cleaning | A clean, well-structured dataset |
| 3. Analyse data | Key metrics and trends | Calculations, comparisons, patterns, outliers |
| 4. Visualise data | Designing charts and dashboards | Charts or a dashboard that answer the questions |
It is a cycle because the first version is rarely right. Building a chart often reveals a data problem (go back to stage 2) or a better question (go back to stage 1).
Start with the problem, then break it into analytical questions that data can answer. "Profit is disappointing" is a problem. Questions could be "Which categories have the lowest profit ratio?", "Is profit lower on orders with higher discounts?" and "Has profitability changed over time?"
Then define information needs: which metrics (Sales, Profit, profit ratio), at what level of detail (order line, customer, month, province), for which period, and for which audience. A director needs a few headline KPIs; an analyst needs detail. Agree on scope early, so you do not chart everything.
Identify the sources (for example, order, customer and product tables), check how they link, and clean the data: fix inconsistent labels, handle missing values and duplicates, and check value ranges. Module 2 covers this in Tableau and Tableau Prep. Poor preparation causes double counting and wrong totals that no chart design can repair.
Compute the agreed metrics, compare groups, look at trends over time, and look for patterns and outliers. Ask whether a pattern is real and large enough to matter, and whether another factor could explain it.
Select a visual method for each question using the chart families from lesson 1.4 (comparison, trend, composition, relationship, distribution, geographic). Build the chart, then check it:
Refine colours, labels, sorting and titles. Interpret what the chart shows in one or two sentences, including limits ("this is correlation, not proof of cause"). Communicate the result in the form the audience needs: a single chart in a report, a dashboard or a data story (Module 5).
Show a draft to a user early. Typical feedback ("I need this by region", "what does this colour mean?") sends you back to an earlier stage. Each loop should make the output more accurate and easier to use.
The sales director of the course's office-supplies retailer says: "Our profit margin feels low. Find out why."
Flashcards · Data Visualization Process
8 wordsA cycle of defining requirements, collecting and preparing data, analysing and visualising, repeated through iteration.
The first stage: set objectives, scope, analytical questions, metrics and audience.
A specific question that available data can answer, derived from a broader business problem.
The metrics, level of detail, period and audience required to answer a question.
The stage of gathering sources and cleaning, standardising and linking them into an analysis-ready dataset.
Returning to earlier stages after checks or feedback to improve accuracy and usefulness.
Verifying totals, scales, labels and design principles before a chart is shared.
A short statement of what a chart shows and its limits, linked to the business question.