plot2llm

Convert your Python plots into LLM-ready structured outputs — from matplotlib and seaborn.

Plot2LLM bridges the gap between data visualization and AI. Instantly extract technical summaries, JSON, or LLM-optimized context from your figures for explainable AI, documentation, or RAG pipelines.

🧠 Use the 'semantic' format to generate structured context optimized for GPT, Claude or any RAG pipeline.

Latest Updates (v0.2.1):

• ✅ Complete Statistical Insights: Full distribution analysis, correlations, outliers, and central tendency for all plot types
• ✅ Enhanced Plot Type Detection: Improved histogram vs bar vs line detection with proper prioritization
• ✅ Rich Pattern Analysis: Detailed shape characteristics and pattern recognition for all visualization types
• ✅ Comprehensive Test Suite: 172/174 tests passing (98.9% success rate) with 24s execution time
• ✅ Production Ready: All core features validated with extensive error handling and edge case coverage
• ✅ Code Quality: All linting issues resolved with ruff and black formatting

---

Features

Feature	Status
Matplotlib plots	✅ Full support
Seaborn plots	✅ Full support
JSON/Text/Semantic output	✅
Custom formatters/analyzers	✅
Multi-axes/subplots	✅
Level of detail control	✅
Error handling	✅
Extensible API	✅
Statistical Analysis	✅ Complete
Pattern Analysis	✅ Rich insights
Axis Type Detection	✅ Smart detection
Unicode Support	✅ Full support
Distribution Analysis	✅ Skewness/Kurtosis
Correlation Analysis	✅ Pearson/Spearman
Outlier Detection	✅ IQR method
Plot Type Detection	✅ Histogram/Bar/Line
Plotly/Bokeh/Altair detection	🚧 Planned
Jupyter plugin	🚧 Planned
Export to Markdown/HTML	🚧 Planned
Image-based plot analysis	🚧 Planned

---

Who is this for?

• Data Scientists who want to document or explain their plots automatically
• AI engineers building RAG or explainable pipelines
• Jupyter Notebook users creating technical visualizations
• Developers generating automated reports with AI
• Researchers needing statistical analysis of visualizations

---

Installation

pip install plot2llm

For full functionality with matplotlib and seaborn:

pip install plot2llm[all]

Note: Version 0.2.1 includes all required dependencies (scipy, jsonschema) for complete functionality.

Or, for local development:

git clone https://github.com/Osc2405/plot2llm.git
cd plot2llm
pip install -e .

---

Quick Start

import matplotlib.pyplot as plt
import numpy as np
from plot2llm import FigureConverter

x = np.linspace(0, 2 * np.pi, 100)
fig, ax = plt.subplots()
ax.plot(x, np.sin(x), label="sin(x)", color="royalblue")
ax.plot(x, np.cos(x), label="cos(x)", color="orange")
ax.set_title('Sine and Cosine Waves')
ax.set_xlabel('Angle [radians]')
ax.set_ylabel('Value')
ax.legend()

converter = FigureConverter()
text_result = converter.convert(fig, 'text')
print(text_result)

---

Examples

Basic Usage

import matplotlib.pyplot as plt
from plot2llm import FigureConverter

fig, ax = plt.subplots()
ax.bar(['A', 'B', 'C'], [10, 20, 15], color='skyblue')
ax.set_title('Bar Example')

converter = FigureConverter()
print(converter.convert(fig, 'text'))

Advanced Statistical Analysis

import seaborn as sns
import matplotlib.pyplot as plt
from plot2llm import FigureConverter

# Create a scatter plot with correlation
fig, ax = plt.subplots()
x = np.random.randn(100)
y = 2 * x + np.random.randn(100) * 0.5
ax.scatter(x, y)
ax.set_title('Correlation Analysis')

converter = FigureConverter()
semantic_result = converter.convert(fig, 'semantic')

# Access statistical insights
stats = semantic_result['statistical_insights']
print(f"Correlation: {stats['correlations'][0]['value']:.3f}")
print(f"Strength: {stats['correlations'][0]['strength']}")

Real-World Examples

The examples/ directory contains comprehensive examples:

• minimal_matplotlib.py: Basic matplotlib usage
• minimal_seaborn.py: Basic seaborn usage
• real_world_analysis.py: Financial, marketing, and customer segmentation analysis
• llm_integration_demo.py: LLM integration and format comparison
• semantic_output_*.py: Complete semantic output examples

Run any example with:

python examples/minimal_matplotlib.py

---

Output Formats

Text Format

Plot types in figure: line
Figure type: matplotlib.Figure
Dimensions (inches): [8.0, 6.0]
Title: Demo Plot
Number of axes: 1
...

JSON Format

{
  "figure_type": "matplotlib",
  "title": "Demo Plot",
  "axes": [...],
  ...
}

Semantic Format (LLM-Optimized)

{
  "metadata": {
    "figure_type": "matplotlib",
    "detail_level": "medium"
  },
  "axes": [
    {
      "title": "Demo Plot",
      "plot_types": [{"type": "line"}],
      "x_type": "numeric",
      "y_type": "numeric"
    }
  ],
  "statistical_insights": {
    "central_tendency": {"mean": 0.5, "median": 0.4},
    "correlations": [{"type": "pearson", "value": 0.95, "strength": "strong"}]
  },
  "pattern_analysis": {
    "pattern_type": "trend",
    "shape_characteristics": {
      "monotonicity": "increasing",
      "smoothness": "smooth"
    }
  }
}

---

Advanced Features

Statistical Analysis

• Central Tendency: Mean, median, mode calculations
• Variability: Standard deviation, variance, range analysis
• Correlations: Pearson correlation coefficients with strength and direction
• Data Quality: Total points, missing values detection
• Distribution Analysis: Skewness and kurtosis for histograms

Pattern Analysis

• Monotonicity: Increasing, decreasing, or mixed trends
• Smoothness: Smooth, piecewise, or discrete patterns
• Symmetry: Symmetric or asymmetric distributions
• Continuity: Continuous or discontinuous data patterns

Smart Axis Detection

• Numeric Detection: Handles Unicode minus signs and various numeric formats
• Categorical Detection: Identifies discrete categories vs continuous ranges
• Mixed Support: Works with both Matplotlib and Seaborn plots

---

API Reference

See the full API Reference for details on all classes and methods.

---

Project Status

This project is in stable beta. Core functionalities are production-ready with comprehensive test coverage.

• Matplotlib support (Full)
• Seaborn support (Full)
• Extensible formatters/analyzers
• Multi-format output (text, json, semantic)
• Statistical analysis with correlations
• Pattern analysis with shape characteristics
• Smart axis type detection
• Unicode support for numeric labels
• Comprehensive error handling
• Plotly/Bokeh/Altair integration
• Jupyter plugin
• Export to Markdown/HTML
• Image-based plot analysis

---

Changelog

v0.2.1 (Latest)

• ✅ Enhanced Statistical Analysis: Complete statistical insights for all plot types
• ✅ Improved Plot Type Detection: Better histogram vs bar vs line detection
• ✅ Rich Pattern Analysis: Detailed shape characteristics for all visualization types
• ✅ Comprehensive Test Suite: 172/174 tests passing (98.9% success rate)

v0.2.1

• ✅ Enhanced Statistical Analysis: Complete statistical insights for all plot types
• ✅ Improved Plot Type Detection: Better histogram vs bar vs line detection
• ✅ Rich Pattern Analysis: Detailed shape characteristics for all visualization types
• ✅ Comprehensive Test Suite: 172/174 tests passing (98.9% success rate)

---

Contributing

Pull requests and issues are welcome! Please see the docs/ folder for API reference and contribution guidelines.

---

License

MIT License

---

Contact & Links

• GitHub repo
• Issues

---

Try it, give feedback, or suggest a formatter you'd like to see!