Chart Selector — Data Visualization Chart Selection Guide
A skill that enhances visualization design for the dashboard-builder.
Target Agents
- dashboard-builder — Selects the optimal chart for the data
- kpi-designer — Determines metric presentation methods
Chart Selection Decision Tree
Question 1: What are you trying to show?
├── Comparison > Question 2a
│ ├── Between items > Horizontal bar / Vertical bar
│ ├── Over time > Line chart / Area chart
│ └── Part-to-whole > Stacked bar / Pie (5 or fewer)
│
├── Trend > Line chart / Area chart
│ ├── Single metric > Line chart
│ ├── Multiple metrics > Multi-line (max 5)
│ └── Range display > Area chart
│
├── Distribution > Histogram / Box plot
│ ├── Single variable > Histogram
│ ├── Group comparison > Box plot
│ └── 2-variable relationship > Scatter plot
│
├── Composition > Pie / Donut / Treemap
│ ├── 2-5 items > Pie/Donut
│ ├── 6+ items > Treemap
│ └── Composition change over time > 100% stacked area
│
└── Relationship > Scatter / Bubble / Heatmap
├── 2 variables > Scatter plot
├── 3 variables > Bubble chart
└── Matrix > Heatmap
Detailed Chart Type Guide
KPI Card
Purpose: Display key metrics at a glance
Suited for: Single number + change rate
Elements: Current value, period-over-period change, trend sparkline
Rules:
- Place at top of dashboard
- 3-6 cards is optimal
- Colors: Increase=green, Decrease=red (reverse for costs)
Line Chart
Purpose: Trends over time
Suited for: Continuous data, time series
Rules:
- Maximum 5 lines (excess: highlight + gray)
- Y-axis starts at 0 (except ratios)
- Hide markers with 10+ data points
- Annotate events (promotions, outages, etc.)
Bar Chart
Purpose: Comparison between items
Suited for: Categorical data
Rules:
- Vertical: Time-axis comparison (monthly, weekly)
- Horizontal: Item comparison (by product, by region) — better label readability
- Sort: By size (except time-based)
- Color: Same color, highlight only with accent
Pie/Donut
Purpose: Part-to-whole ratios
Suited for: 2-5 categories
Prohibited:
- 6+ items > Use treemap/horizontal bar
- Similar proportions > Incomparable, use bar
- 3D pie > Never use
Rules:
- Start at 12 o'clock, go clockwise
- Largest item first
- "Other" item last
Heatmap
Purpose: Express intensity in a 2D matrix
Suited for: Time x category, correlation matrix
Rules:
- Single color gradient (sequential)
- Diverging data: 2 colors from center
- Display values in cells (ensure readability)
Table
Purpose: Provide exact figures
Suited for: Multiple measures, detailed comparison
Rules:
- Sortable columns
- Conditional formatting (highlight top/bottom)
- Optional sparkline column
- 10-20 rows optimal (pagination)
Color Palette Guide
Sequential — Size/intensity
Low ──────────────── High
#F7FBFF > #6BAED6 > #08306B (blue)
Diverging — Bidirectional from center
Negative ──── 0 ──── Positive
#D73027 > #FFFFBF > #1A9850 (red-yellow-green)
Categorical — Distinction
Maximum 8 colors: supplement with patterns/labels if more needed
Accessibility: Use colorblind-safe palettes (viridis, ColorBrewer)
Dashboard Layout Principles
1. F-pattern: Left to right, top to bottom reading flow
2. Priority order: Key KPIs at top, details at bottom
3. Grouping: Place related charts close together
4. Whitespace: Sufficient spacing between charts
5. Responsive: 2 columns mobile, 3-4 columns desktop
6. Interaction: Filter > drill-down > detail sequence
Layout grid:
+----------+----------+----------+----------+
| KPI 1 | KPI 2 | KPI 3 | KPI 4 | <- Summary
+----------+----+-----+----------+----------+
| Trend Chart | Comparison | Composition| <- Analysis
+---------------+--------------+------------+
| Detail Table | <- Detail
+-------------------------------------------+