MCP Server For Data Analysis

🚀 MCP Data Analysis Server

This server is primarily used for integrating data analysis tools. It also includes some resource examples and prompt information. Use FastMCP to define and run the server, and expose each tool as an endpoint.

✨ Features

• Read Excel files from the desktop.
• Perform descriptive analysis.
• Create charts (e.g., bar charts, pie charts, line charts, scatter plots, histograms, and box plots).

🛠️ Tools

List Excel Files

files_list_excel_files(): Scan the desktop for .xlsx and .xls files and return a list of file names.

Read an Excel File

read_excel_file(filename, sheet_name=None): Read the specified Excel file from the desktop. If sheet_name is not provided, it reads the first worksheet by default. Returns basic information about the number of rows, number of columns, a view of the first five rows of data, and data types.

Analyze Excel Data

analyze_excel_data(filename, sheet_name=None): Calculate descriptive statistics for numerical, categorical, and date columns. If applicable, it also calculates the correlation matrix for numerical columns.

Chart Generation

These tools read data from Excel files and use Matplotlib to generate charts. The charts are returned as Image objects (PNG data in memory), which can be processed by the client later.

• create_bar_chart(filename, x_column, y_column, sheet_name=None, title=None, limit=10)
• create_pie_chart(filename, labels_column, values_column, sheet_name=None, title=None, limit=8)
• create_line_chart(filename, x_column, y_columns, sheet_name=None, title=None)
• create_scatter_plot(filename, x_column, y_column, color_column=None, sheet_name=None, title=None)
• create_histogram(filename, column, bins=10, sheet_name=None, title=None)
• create_box_plot(filename, columns, sheet_name=None, title=None)

🖥️ MCP Server

When the script is run directly (i.e., python excel_analytics_server.py), it starts the FastMCP server and registers all the above tools. These tools can be called by clients that support the MCP interface.

Using Claude Desktop

Prerequisites

• Download the Claude desktop version.

Step 1: Add MCP to a Python Project

# Install uv
curl -LsSf https://astral.sh/uv/install.sh | sh

# Create a new project directory
uv init data_analysis_tool
cd data_analysis_tool

# Create and activate a virtual environment
uv venv
source .venv/bin/activate

# Install dependencies
uv add "mcp[cli]" httpx

Step 2: Create an MCP Server

• Place the file excel_analytics_server.py in the data_analysis_tool directory.
• Run the following command:

pip install "mcp[cli]" pandas matplotlib openpyxl

Step 3: Test the Server

Run the following command to start the server:

python excel_analytics_server.py

Open a browser and visit http://localhost:8000 to view the available interfaces.

Step 4: Connect to Claude Desktop

In the Claude desktop, navigate to "File" -> "New Prompt" and enter the following to test the server:

/Docker

📚 Resources and Tips

• For more detailed information, refer to the official documentation.
• Example image: