Data Visualization Types
Common chart types and what they work for — when to use bar, line, scatter, box, heatmap, etc.
Reference
By question type
| Goal | Best chart | Notes |
|---|---|---|
| Compare values across categories | Bar / column chart | Order bars by value, not alphabetically |
| Trend over time | Line chart | X = time, Y = metric |
| Part of a whole | Stacked bar (not pie) | Pie charts are hard to read past 3 slices |
| Distribution of one variable | Histogram / box plot / violin | Box for summary, violin for shape |
| Relationship between two variables | Scatter plot | Add a regression line if helpful |
| Three variables | Bubble / 3D scatter / faceted | Encoding: x, y, size/color |
| Proportions over time | Streamgraph / stacked area | Good when total also matters |
| Matrix of correlations | Heatmap | Use a diverging palette centered on 0 |
| Geographic data | Choropleth / bubble map | Normalize by population if counting |
| Hierarchical parts | Treemap / sunburst | Good for nested categories |
| Flow between states | Sankey diagram | Watch for clutter with many categories |
| Ranked comparisons over time | Slope chart / bump chart | Shows rank changes |
Anti-patterns
- Pie chart with 12+ slices — use a bar chart.
- 3D bar chart — adds no info, distorts comparison.
- Dual Y-axis with unrelated scales — invites misreading. Prefer two stacked panels.
- Truncated Y-axis that starts above zero on a bar chart — exaggerates differences.
- Rainbow palette for ordered data — use a monotone sequential scale.
- Red/green only — excludes ~5% of viewers with color blindness.
Last updated: