Why this matters
As a Data Visualization Engineer, choosing the right chart is half the insight. Real tasks you will do include: selecting visuals for dashboards, guiding stakeholders through trends and comparisons, turning messy metrics into clear decisions, and preventing confusion caused by poor chart choices. Good selection improves comprehension speed, reduces errors, and helps decisions happen faster.
Who this is for
- Data Visualization Engineers building dashboards and reports
- Analytics and BI professionals who translate analysis into visuals
- Anyone who wants to communicate data clearly without guesswork
Prerequisites
- Basic understanding of measures vs. dimensions
- Comfort reading line, bar, scatter, and pie charts
- Familiarity with time series, categories, and continuous variables
Concept explained simply
Chart selection is mapping your message to a visual form your audience decodes quickly and accurately.
Mental model
- Message: What should the viewer learn? (trend, comparison, relationship, distribution, part-to-whole, ranking, flow, location)
- Data shape: Time series? Categories? Continuous values? Single vs. multiple series?
- Precision: Do we need exact comparisons or overall feel?
- Audience: Technical or non-technical? What’s familiar to them?
- Constraints: Space, small screens, color limitations, accessibility.
Open: Quick mapping from message to chart family
- Comparison (few categories): Bar/column (sort descending)
- Trend over time: Line (single/multi-series), Area (show total/composition)
- Distribution: Histogram, Box plot (for medians/outliers)
- Part-to-whole (few parts): Donut/Pie (sparingly); better: Bar with % of total
- Composition over time: Stacked area or 100% stacked bars
- Relationship (2 vars): Scatter; add trendline
- Ranking (many categories): Horizontal bars, pareto
- Flow: Sankey/Alluvial (when movement matters)
- Geospatial: Choropleth for rates; symbol map for counts
Tip: If exact comparisons matter, prefer bars/lines over pies/areas.
Step-by-step selection process
- Identify the analytical task: trend, comparison, distribution, relationship, composition, ranking, flow, map.
- Count series and categories: Many categories? Prefer horizontal bars or small multiples.
- Decide on precision: Need exact? Use bars and aligned baselines. Need overall pattern? Lines/areas are fine.
- Consider accessibility: Avoid relying on color alone; label directly when possible.
- Stress test for clutter: If lines cross too much, use small multiples or filter to top N.
- Annotate the key point: Call out peak, change, or threshold to focus attention.
Worked examples
Example 1: Quarterly revenue by region across 6 quarters
Goal: Show trend and compare regions. Choose a multi-series line chart (one line per region). If lines overlap, use small multiples (one panel per region) sharing axes. Annotate the quarter with a major change.
Example 2: Budget split across 8 categories
Goal: Compare parts precisely. Avoid a pie with 8 slices. Use a sorted horizontal bar chart with each bar labeled with % of total. If the audience needs part-to-whole feel only, a donut with 3–5 categories may be acceptable.
Example 3: Delivery time variability by service A vs B
Goal: Compare distributions and outliers. Use side-by-side box plots or mirrored histograms. Box plots communicate medians, IQR, and outliers quickly.
Example 4: Marketing spend vs sign-ups across 100 campaigns
Goal: Relationship between two continuous variables. Use a scatter plot; add a trend line and optionally encode campaign type by color or shape (ensure accessible contrast).
Common mistakes and how to self-check
- Pies with too many slices: If categories > 5 or differences are slight, use bars.
- Bars without zero baseline: Bars must start at zero or comparisons mislead.
- Overloaded multi-series lines: If more than ~5 lines, use small multiples or highlight a focus line.
- Using choropleth for counts: Normalize by population/area; otherwise big regions dominate visually.
- Encoding too many variables with color: Use position/length first; color only when necessary.
Self-check: Can a new viewer state the main message in 5 seconds? Can they rank the top 3? Do annotations match the insight?
Exercises
Do these now; they mirror the graded quick test.
- Exercise 1: Choose a chart for four scenarios (trend, distribution, part-to-whole, relationship). Justify briefly.
- Exercise 2: Redesign a misleading 3D pie into clear charts; explain encoding and annotations.
- Checklist before submitting:
- Chart matches the analytical task
- Clutter reduced (top N or small multiples)
- Labels/annotations clarify the takeaway
- Axes and baselines are appropriate
Practical projects
- Redesign a company dashboard: Replace 3 unclear charts with alternatives and write 1-sentence justifications for each choice.
- Small-multiples story: Take a time series split by 8+ categories and build small multiples with a shared y-axis and one key annotation per panel.
- Map vs bars: Show a rate by region as a choropleth and as a bar chart; write which is better for ranking and why.
Learning path
- Start: Chart Type Selection (this lesson)
- Next: Visual Encoding & Perception basics
- Then: Dashboard layout and annotation
- Optional: Color theory and accessibility
Next steps
- Finish the exercises below and compare with the provided solutions.
- Take the Quick Test to check mastery.
- Apply one project in your current work or a portfolio piece.
Progress and test
You can take the quick test now. Everyone can take it for free; progress is saved only for logged-in users.
Mini challenge
Pick one dataset you know well (e.g., weekly traffic by channel). Create two versions: a multi-series line and small multiples. Ask a peer which one helps them answer the intended question faster. Note their time-to-answer and preference. Choose the winner and write down the reason.