mcpField
P
plot-from-data

plot-from-data

github.com/trae1oungData & databases91965 installsUpdated 27w ago
Repository

About plot-from-data

Plot From Data is a skill for filling paper-style figure templates with user data. It provides scripts for bar, line, scatter, and radar charts, with output as 300 dpi PNG files.

Key features

  1. 01Provides eight figure style templates
  2. 02Supports bar, line, scatter, and radar charts
  3. 03Fills data by editing script data sections
  4. 04Reference files specify exact style parameters
  5. 05Outputs 300 dpi PNG files

Use cases

  1. 01Plotting paired baseline and method comparisons
  2. 02Showing ablations or training curves
  3. 03Presenting t-SNE clustering results
  4. 04Comparing methods across multiple dimensions

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/trae1oung/paper-plot-skills --skill plot-from-data

After installing, type/plot-from-datain Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

agents/
openai.yaml
references/
bar_grouped_hatch.md
bar_paired_delta.md
line_confidence_band.md
line_loss_with_inset.md
line_training_curve.md
radar_dual_series.md
scatter_broken_axis.md
scatter_tsne_cluster.md
scripts/
bar_memevolve.py
bar_spice.py
line_aime.py
line_loss_inset.py
line_selfdistill.py
radar_dora.py
scatter_break.py
scatter_tsne.py

Details

From the project README

Plot From Data

Generate a paper-quality figure by picking a style template and filling it with user data. All outputs are dpi=300 PNG.

Available Styles

StyleTypeScript适用场景
bar_paired_delta柱状图scripts/bar_memevolve.pyBaseline vs method 配对对比 + 增益箭头
bar_grouped_hatch柱状图scripts/bar_spice.py多方法消融，主方法斜线填充，柱顶数值
line_confidence_band折线图scripts/line_selfdistill.py带置信区间的训练曲线
line_training_curve折线图scripts/line_aime.py垂直断点线 + 水平参考线
line_loss_with_inset折线图scripts/line_loss_inset.pyL 形 spine + 局部放大 inset
scatter_tsne_cluster散点图scripts/scatter_tsne.pyt-SNE 聚类 + 注释框
scatter_broken_axis散点图scripts/scatter_break.py折断 X 轴，多 marker 系列
radar_dual_series雷达图scripts/radar_dora.py双方法多维对比，正八边形网格

Workflow

1. 确认用户的图类型和数据
2. 选择对应 style（如不确定，询问用户或根据数据形状推断）
3. 读取对应 references/<style_name>.md 获取精确参数
4. 复制对应 scripts/<script>.py，替换数据区（脚本顶部有清晰注释标注数据区）
5. 运行：python3 scripts/<script>.py
6. 检查输出，必要时微调颜色/标签/字号

Data Substitution Tips

每个 repro 脚本的数据区在文件顶部，通常是 np.array(...) 或字典。替换规则：

  • 保持数组维度和类型不变
  • 若类别数变化（如从 4 组改为 6 组），同步调整颜色列表和宽度计算
  • x 轴标签、图例标签直接修改对应字符串列表

Detailed Style Parameters

Read the corresponding file in references/ for exact rcParams, colors, font sizes, spine settings, and tick directions before generating:

  • Bar: references/bar_paired_delta.md, references/bar_grouped_hatch.md
  • Line: references/line_confidence_band.md, references/line_training_curve.md, references/line_loss_with_inset.md
  • Scatter: references/scatter_tsne_cluster.md, references/scatter_broken_axis.md
  • Radar: references/radar_dual_series.md

Comments

Markdown supported