promptadvisers/forecasting-calculator-mcp

Forecasting Calculator MCP Server

A sophisticated Model Context Protocol (MCP) server that provides multiple statistical forecasting methods for time series data.

Features

• Multiple Forecasting Methods:

  • Linear Regression - For steady trends
  • Moving Average - For smoothing fluctuations
  • Exponential Smoothing - For recent-weighted predictions
  • Polynomial Regression - For non-linear patterns
  • Simple ARIMA - For autocorrelated data

• Interactive Usage: Ask for forecasts with natural language

• Mathematical Explanations: Get detailed breakdowns of calculations

• Flexible Input: Accepts arrays or space/comma-separated numbers

• Configurable Periods: Forecast 1-100 periods into the future

Installation

1. Create and activate virtual environment:

cd forecasting-calculator-mcp
python3 -m venv mcp_venv
source mcp_venv/bin/activate

2. Install dependencies:

pip install -r requirements.txt

3. Make scripts executable:

chmod +x forecasting_calculator_server.py
chmod +x run_forecasting_calculator.sh

Testing

Test the server directly:

# Test initialization
echo '{"jsonrpc": "2.0", "method": "initialize", "params": {"protocolVersion": "2025-06-18", "capabilities": {}, "clientInfo": {"name": "test", "version": "1.0"}}, "id": 1}' | python forecasting_calculator_server.py

# Test tool listing
echo '{"jsonrpc": "2.0", "method": "tools/list", "params": {}, "id": 2}' | python forecasting_calculator_server.py

Claude Desktop Configuration

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "forecasting-calculator": {
      "command": "/path/to/your/forecasting-calculator-mcp/run_forecasting_calculator.sh",
      "args": [],
      "env": {}
    }
  }
}

Note: Replace /path/to/your/ with the actual path where you installed the server.

Usage Examples

In Claude Desktop:

1. Simple Linear Forecast:

   • "Forecast [300, 400, 500] for 3 periods using linear regression"

2. Revenue Projection:

   • "My revenue was 1000, 1200, 1400, 1650. Forecast next 6 months with exponential smoothing"

3. Understand Methods:

   • "Explain the forecasting methods available"

Methods Overview

Troubleshooting

Technical Details

• Uses standard MCP SDK (not FastMCP)
• Includes ServerCapabilities in initialization
• Safe mathematical operations with NumPy
• Comprehensive error handling
• Virtual environment isolation

License

MIT