Why this matters
Choosing the wrong chart hides insights or misleads decisions. As a Data Analyst, you will:
- Present KPI trends to executives (choose legible time-series visuals).
- Compare segments and rankings (ensure fair comparisons and readable labels).
- Explain distributions and outliers (pick charts that show spread clearly).
- Show part-to-whole composition (avoid clutter, highlight key shares).
- Reveal relationships (use encodings that make correlations visible).
Good chart selection speeds decisions, reduces back-and-forth, and builds trust.
Concept explained simply
Charts are tools to answer a specific question. First decide the message, then pick the chart that makes that message obvious.
Mental model: Message → Data type → Chart
- Message: trend, compare, rank, part-to-whole, distribution, relationship, composition over time.
- Data type: categorical (nominal/ordinal), time (ordered), numeric (continuous).
- Chart: choose the one that maps data type to your message with the clearest visual encoding.
Use the strongest encodings first: position on a common scale > length > angle > area > color hue.
Quick chooser (most common cases)
- Trend over time: line chart (few series), area chart (emphasize volume), small multiples for many series.
- Compare categories: bar chart (horizontal for long labels, sort by value).
- Rank (best to worst): sorted bar chart.
- Part-to-whole (few categories): pie or donut is acceptable; for exact comparison use stacked or separate bars.
- Distribution of one variable: histogram, box plot (for medians/outliers), violin (advanced).
- Relationship between two numeric variables: scatter plot (+ trend line if appropriate).
- Composition over time: stacked area (absolute), 100% stacked area (share).
- Many categories (20+): bar chart with filtering/aggregation; avoid pies/treemaps for precise comparison.
Worked examples
Example 1: Monthly revenue for the last 24 months
Goal: show the trend and seasonality.
Best choice: line chart with monthly points; optionally add a 3-month moving average.
Why not other charts?
- Pie: no natural order; can’t show trend.
- Bar: okay, but line shows continuity and seasonal pattern more clearly.
Example 2: Market share of 5 brands this quarter
Goal: show contributions and allow quick comparison.
Best choice: sorted bar chart with total = 100% noted. If audience prefers part-to-whole at a glance, a simple donut with labels can work, but bars compare differences more precisely.
Why not stacked bars?
Single period stacked bar is fine, but sorting bars individually gives cleaner comparisons across brands.
Example 3: Customer satisfaction scores (0–10) from 5,000 responses
Goal: show distribution shape and outliers.
Best choice: histogram to show skew; add a box plot summary (median, quartiles) below.
Why not average in a single number?
Averages hide skew and multimodality; distribution reveals true spread.
Example 4: Does higher traffic correlate with higher conversion?
Goal: show relationship between two numeric variables.
Best choice: scatter plot with a fitted trend line and correlation value; encode segment with color if needed (max 4–6 categories).
Why not dual-axis line chart?
Dual axes can mislead due to scale manipulation. Scatter isolates the relationship directly.
Step-by-step selection process
Common mistakes and self-check
- Using pies with many slices (6+). Self-check: can you rank slices quickly? If not, use bars.
- Dual y-axes that imply false relationships. Self-check: could identical trends be an artifact of scaling?
- Too many series in one chart. Self-check: more than 4–6 lines? Use small multiples or filtering.
- Unsorted categories. Self-check: are bars sorted by value or logical order?
- Color misuse (rainbow, low contrast). Self-check: simulate grayscale mentally; would it still work?
- Wrong baseline (non-zero bar chart baseline). Self-check: bar charts should start at zero.
- Label overload. Self-check: can you replace gridlines with direct labels or use light gridlines?
Quick self-audit checklist
- Message clear in title/subtitle.
- Chart matches message and data type.
- Scales appropriate (zero baseline for bars; log scale only for wide positive ranges).
- Colors minimal and meaningful; accessible palette.
- Annotations highlight the key insight.
Practical projects
- Build a KPI trend dashboard: 3–5 monthly metrics as small multiples of line charts; annotate key events.
- Category comparison report: top 15 products as a sorted horizontal bar chart; add filters for region and quarter.
- Distribution deep-dive: histogram + box plot for delivery times; recommend process improvements based on spread and outliers.
Exercises
Do these exercises, then compare with the solutions provided in the exercise cards below.
- Match chart to goal (3 scenarios): Choose the best chart, specify sorting, labels, and any helpful annotations.
- Fix misleading visuals (2 cases): Identify the issue and propose a better chart and scale choice.
Exercise checklist
- Stated the message for each scenario.
- Identified data types and constraints (count of categories/series).
- Selected a chart with a clear reason.
- Specified sorting, scales, and labeling choices.
- Addressed accessibility and readability.
Quick test
Take the short test to check your understanding. Everyone can take it; only logged‑in users will have progress saved automatically.
Learning path
- Next: Color, contrast, and accessibility basics.
- Then: Annotating charts and storytelling structure.
- Finally: Dashboard layout and interaction patterns.
Who this is for
- Aspiring and junior Data Analysts.
- Professionals who present data to stakeholders.
- Anyone refining dashboard/report design skills.
Prerequisites
- Basic understanding of data types (numeric, categorical, time).
- Comfort with a visualization tool (e.g., spreadsheets or BI tools).
- Basic statistics concepts (mean, median, distribution).
Mini challenge
You have a dataset with daily website sessions, sign-ups, and channel (Paid, Organic, Referral) for the last 180 days. Create one page with:
- A line chart for sign-ups over time with a 7-day moving average.
- A 100% stacked area chart showing sign-up share by channel over time.
- A scatter plot of sessions vs sign-ups colored by channel.
Tips
- Annotate campaigns and outages.
- Limit colors to the three channels plus a neutral benchmark.
- Check that axes are consistent and readable.
Next steps
- Apply the chooser to your current report and replace one unclear chart.
- Ask a peer to explain your main message in 5 seconds after viewing your chart.
- Proceed to the quick test to lock in the concepts.