EDA MCP Server - Python

A simple Exploratory Data Analysis (EDA) server that analyzes your CSV and Excel files. Just drop your data files in the folder and get instant insights with beautiful tables and visualizations.

🎯 What This Does

• Analyzes your data files - Statistics, correlations, missing values, distributions
• Creates visualizations - Histograms, scatter plots, correlation heatmaps
• Shows results in tables - Clean, professional ASCII tables
• Works with CSV/Excel - Just drop files in the folder
• Fast and cached - Analyzes large datasets efficiently

🚀 Quick Start

1. Setup (First Time Only)

🚀 This repo includes a pre-built virtual environment for instant use!

# On macOS/Linux - Just activate the included environment
source .venv/bin/activate

# Test installation (should work immediately!)
uv run client_test.py

📝 For Windows users or if you prefer your own environment:

# Create your own virtual environment
uv venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate
uv sync

# Test installation
uv run client_test.py

2. Add Your Data

Drop your CSV or Excel files in this folder:

hw3 - mcp-server/
├── your_data.csv     ← Your files here
├── sales_data.xlsx   ← Excel files work too
├── main.py
└── client_test.py

3. Run Analysis

# Activate environment (if not already active)
source .venv/bin/activate

# Run full analysis demo
uv run client_test.py

📊 What You'll See

When you run the analysis, you get:

1. Quick Data Preview

📊 Dataset: sales_data.csv
📏 Shape: 1,000 rows × 6 columns  
📋 Columns: age, income, education_years, satisfaction_score, department, has_missing
✅ No missing values

2. Statistical Summary Table

+--------------------+---------+----------+----------+--------+----------+----------+----------+-----------+
| Column             |   Count |     Mean |      Std |    Min |      25% |      50% |      75% |       Max |
+====================+=========+==========+==========+========+==========+==========+==========+===========+
| age                |    1000 |    34.67 |     9.8  |   2    |    28    |    35    |    41    |     73    |
| income             |    1000 | 50399.3  | 50149.4  | 161.17 | 14197.7  | 36296.6  | 68584.2  | 372086    |
+--------------------+---------+----------+----------+--------+----------+----------+----------+-----------+

3. Data Overview

📁 File Information:
------------  ---------------
File Path     sales_data.csv
Rows          1,000
Columns       6
Memory Usage  0.09 MB
------------  ---------------

📋 Column Types:
+-------------+---------+--------------------------------+
| Type        |   Count | Columns                        |
+=============+=========+================================+
| Numerical   |       5 | age, income, education_years   |
| Categorical |       1 | department                     |
+-------------+---------+--------------------------------+

4. Missing Data Analysis

📊 Missing Data Analysis (Total rows: 1,000)

+-------------+-----------------+--------------+
| Column      |   Missing Count | Percentage   |
+=============+=================+==============+
| has_missing |             108 | 10.8%        |
+-------------+-----------------+--------------+

🔧 Available Analysis Tools

The server provides these analysis tools:

Tool	What It Does	Example Use
statistical_summary	Mean, std, min/max for numbers	Understanding data distribution
data_overview	File info, column types, data quality	Getting familiar with dataset
missing_data_analysis	Find and visualize missing values	Data cleaning preparation
correlation_analysis	Correlation matrix with heatmap	Finding relationships between variables
distribution_analysis	Histograms, box plots for a column	Understanding individual variable distribution
create_scatter_plot	X vs Y scatter plot with trend line	Exploring relationships between two variables
list_available_files	Show all CSV/Excel files in folder	File discovery

💡 Usage Tips

For Different File Types

# Works with any CSV file
your_data.csv

# Works with Excel files  
your_data.xlsx
your_data.xls

Quick Tests

# Test all tools quickly
uv run client_test.py --quick

# Full demo with sample data (if no files present)
uv run client_test.py

Adding New Files

1. Drop new CSV/Excel files in the main folder
2. Run uv run client_test.py again
3. It automatically detects and analyzes new files

🛠️ Development

Activate Environment

source .venv/bin/activate

Run the MCP Server Directly

python main.py

Project Structure

hw3 - mcp-server/
├── main.py           # MCP server with analysis tools
├── client_test.py    # Demo client that shows results
├── pyproject.toml    # Dependencies
├── uv.lock          # Locked dependencies
├── .venv/           # Virtual environment
└── your_data.csv    # Your data files here

❓ Troubleshooting

"No module named..." errors

# Make sure you're in the virtual environment
source .venv/bin/activate

# If using the included .venv and it doesn't work on your system:
rm -rf .venv
uv venv
uv sync

"No data files found"

• Make sure your CSV/Excel files are in the main project folder
• Check file extensions: .csv, .xlsx, .xls are supported
• The demo creates sample data if no files are found

Performance issues

• Large files (>100MB) may take longer to process
• The server caches data for faster repeated analysis
• Use clear_dataset_cache tool if memory becomes an issue

💡 Why We Include .venv

This repository includes a pre-built virtual environment to prioritize immediate usability:

• ✅ Clone and run in seconds (no setup required)
• ✅ Perfect for demos, workshops, and quick testing
• ✅ All dependencies pre-installed and tested

Note: The included .venv is built for macOS/Linux. Windows users or those preferring fresh environments can easily create their own using the commands above.

🎯 Next Steps

1. Try with your own data - Drop CSV/Excel files and run analysis
2. Explore different analyses - Each tool shows different insights
3. Integrate with MCP clients - Use with Claude Desktop or other MCP-compatible tools
4. Extend functionality - Add your own analysis tools to main.py

---

Ready to analyze your data? Just run uv run client_test.py and see what insights you discover! 📊✨