This skill guides the use of Python to generate data charts at the level of scientific research papers.
Checklist
- Confirm that the conda environment is activated (research)
- Confirm the chart type and data
- Record the data manifest
- If mock/synthetic data is used, clearly label it as planning data
- Use top-journal color schemes
- Set the resolution to 450 DPI
- Output both PNG and SVG
- Check Chinese font display
- Save to the figures/ directory
1. Environment Requirements
1.1 conda Environment
Default environment name：research
Activation command：
conda activate research
Required libraries：
pip install matplotlib seaborn numpy pandas
If the environment is not configured, call the environment-setup skill.
2. Chart Specifications
2.0 Data Manifest and Mock Data Boundaries
Any data chart must first have a data file and a data manifest. Default path:
figures/data-manifest.md
figures/data/<figure-name>.csv
figures/<section>/<figure-name>.py
figures/<section>/<figure-name>.png
figures/<section>/<figure-name>.svg
figures/data-manifest.md At minimum, record:
| Figure | Data file | Real/mock | Source | Script | Outputs |
|---|
Mock or synthetic data may only be used for planning-version charts. The file name must start with mock_ or synthetic_ , and [待真实实验替换]must be retained in the chart, table, or section draft. Mock data must not be written as "experimental results show".
2.1 Resolution Requirements
| Purpose | DPI | Description |
|---|---|---|
| Journal submission | 300-600 | Required by most journals |
| Top-journal submission | 450+ | Nature/Science, etc. |
| Screen display | 150 | PPT/web pages |
This skill uses 450 DPI by default
2.2 Output Formats
Each chart outputs two formats at the same time:
- PNG: bitmap, suitable for web pages and PPT
- SVG: vector graphic, suitable for journal submission
2.3 Chart Dimensions
| Type | Width (inches) | Applicable scenario |
|---|---|---|
| Single-column figure | 3.5 | Journal single column |
| Double-column figure | 7.0 | Journal double column/full width |
| PPT figure | 10.0 | Presentation slides |
3. Top-Journal Color Schemes
3.1 Nature/Science Style
NATURE_COLORS = ['#2E86AB', '#A23B72', '#F18F01', '#C73E1D', '#95C623']
3.2 Cell Style
CELL_COLORS = ['#4E79A7', '#F28E2B', '#E15759', '#76B7B2', '#59A14F', '#EDC948']
3.3 Colorblind-Friendly Color Scheme
COLORBLIND_SAFE = ['#0077BB', '#33BBEE', '#009988', '#EE7733', '#CC3311', '#EE3377']
3.4 Color Scheme Principles
- ❌ Do not use matplotlib default colors
- ❌ Do not use basic colors such as pure red, pure blue, or pure green
- ✅ Keep the number of colors in the same chart to no more than 5
- ✅ Ensure colorblind friendliness
4. Code Template
"""
Figure X: [图表标题]
论文章节: [所属章节]
"""
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.font_manager as fm
from pathlib import Path
# 中文字体配置
CHINESE_FONT = None
font_candidates = [
'/System/Library/Fonts/STHeiti Light.ttc',
'/System/Library/Fonts/PingFang.ttc',
]
for fp in font_candidates:
if Path(fp).exists():
CHINESE_FONT = fm.FontProperties(fname=fp)
break
plt.rcParams['axes.unicode_minus'] = False
# 顶刊配色
COLORS = ['#4E79A7', '#F28E2B', '#E15759', '#76B7B2', '#59A14F']
def setup_plot_style():
plt.rcParams.update({
'font.size': 10,
'axes.titlesize': 12,
'axes.labelsize': 10,
'axes.spines.top': False,
'axes.spines.right': False,
'axes.grid': True,
'grid.alpha': 0.3,
'legend.frameon': False,
'savefig.dpi': 450,
'savefig.bbox': 'tight',
})
def main():
setup_plot_style()
fig, ax = plt.subplots(figsize=(7, 5))
# === 绑定代码 ===
x = np.linspace(0, 10, 100)
ax.plot(x, np.sin(x), color=COLORS[0], label='Model A')
ax.plot(x, np.cos(x), color=COLORS[1], label='Model B')
if CHINESE_FONT:
ax.set_xlabel('时间 (s)', fontproperties=CHINESE_FONT)
ax.set_ylabel('幅值', fontproperties=CHINESE_FONT)
else:
ax.set_xlabel('Time (s)')
ax.set_ylabel('Amplitude')
ax.legend()
# === 绑定代码结束 ===
# 保存
output_dir = Path(__file__).parent
fig_name = Path(__file__).stem
plt.savefig(output_dir / f'{fig_name}.png', dpi=450)
plt.savefig(output_dir / f'{fig_name}.svg')
plt.show()
if __name__ == '__main__':
main()
5. Common Chart Types
Line chart
ax.plot(x, y, color=COLORS[0], linewidth=1.5, marker='o', markersize=4)
Bar chart
ax.bar(x_pos, values, color=COLORS[:len(values)], edgecolor='white')
Heatmap
im = ax.imshow(matrix, cmap='RdBu_r', aspect='auto')
plt.colorbar(im, ax=ax)
Box plot
bp = ax.boxplot(data_list, patch_artist=True)
for patch, color in zip(bp['boxes'], COLORS):
patch.set_facecolor(color)
Scatter plot
ax.scatter(x, y, c=colors, s=sizes, alpha=0.6, cmap='viridis')
6. File Management
Directory structure
figures/
├── chapter1/
│ ├── fig1_overview.py
│ ├── fig1_overview.png
│ └── fig1_overview.svg
├── chapter2/
└── chapter3/
Naming conventions
- File name format:
fig{序号}_{描述}.py - Example:
fig1_model_architecture.py
7. Quality Check
Chart content
- Data is accurate and correct
- Axis labels are complete (including units)
- Legend is clear and readable
Visual effect
- Use top-journal color schemes
- Resolution reaches 450 DPI
- Font size is moderate
File output
- PNG format has been generated
- SVG format generated
- File naming conventions
VIII. Common Issues
Q1: Chinese characters display as squares
from matplotlib.font_manager import FontProperties
font = FontProperties(fname='/System/Library/Fonts/STHeiti Light.ttc')
ax.set_xlabel('中文标签', fontproperties=font)
Q2: Images are blurry
plt.savefig('figure.png', dpi=450, bbox_inches='tight')
Q3: Legend obscures data
ax.legend(loc='upper left', bbox_to_anchor=(1.02, 1))