📘

Data Visualization

Show the pattern, not the chart junk.

v1 data-science curated
🤖

Install 安装

curl -sSL https://updating.cc/skills/data/data-visualization/SKILL.md -o ~/.agents/skills/data__data-visualization/SKILL.md

SKILL.md Preview 技能内容预览

---
name: Data Visualization
description: Show the pattern, not the chart junk.
---
# Data Visualization

A chart's job is to make a pattern visible at a glance. Decoration (3D, gradients, gratuitous color) obscures the pattern it's supposed to reveal.

## Principles
- **Pick the chart by the question.** Comparison over time → line; comparison across categories → bar; part-of-whole → stacked bar (rarely pie); distribution → histogram/box.
- **Strip chart junk.** Drop gridlines, borders, and gradients that don't encode data. Every pixel should earn its place.
- **Direct label over legend** when there are few series. Eyes jump between legend and bar; a label on the bar removes the trip.
- **Start the y-axis at zero for bar charts.** Truncated axes exaggerate differences and mislead.
- **Color for meaning.** One accent for the story; gray for context. Rainbow color scales on ordinal data are noise.