convez.ai
MCP ServerDados & AnalyticsIntermediário
Varu4

ProData AI — Servidor MCP para Análise de Dados e AutoML

Varu4/prodata-ai-mcp

Análise de dados e AutoML via Claude — sem código, só CSV

Ver no GitHub
InicialPluginScore · de 0 a 100

Servidor MCP que integra análise de dados profissional, AutoML, previsão de séries temporais e geração de relatórios diretamente no Claude, Cursor e VS Code. Processa CSVs via linguagem natural, identifica padrões, treina modelos automaticamente e retorna insights com visualizações.

Instalação

$npx convez-ai Varu4/prodata-ai-mcp

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 5 meses
Versão
—
Issues abertas
2
Contribuidores
1

Casos de uso

  • Análise exploratória de datasets em linguagem natural
  • Treinamento automático de modelos ML com comparação de performance
  • Identificação de features mais importantes para prognóstico
  • Previsão de séries temporais com algoritmos avançados
  • Geração de relatórios executivos com estatísticas e recomendações

Veredito

Ferramenta madura e bem integrada ao ecossistema Claude. Reduz significativamente a barreira técnica para análise de dados profissional, com AutoML eficaz (R² 0.98+ reportado). Requer assinatura MCPize para uso prático (tier gratuito: 50 req/mês), mas instalação é trivial via one-click.

README

A tradução para o português deste README ainda está na fila.

🤖 ProData AI — MCP Server for Data Analysis & AutoML

ProData AI Railway MCPize Python License

Professional data analysis & AutoML — directly inside Claude, Cursor, VS Code, and Windsurf.

🚀 Install via MCPize · 🎮 Try Live Demo · 🛒 Get on Gumroad · 📺 Watch Demo


✨ What is ProData AI?

ProData AI is a production-ready Model Context Protocol (MCP) server that brings professional-grade data analysis, AutoML training, time series forecasting, and full report generation directly into your AI workflow — no code required.

Just upload a CSV and ask Claude (or any MCP-compatible client) to analyze it. ProData AI handles the rest.

🏆 Gradient Boosting achieved R² = 0.9866 on a retail sales dataset, with marketing spend identified as the #1 revenue driver — all via natural language.


🛠️ Tools (5 MCP Tools)

ToolDescription
analyze_dataset_toolFull statistical analysis — mean, median, std, missing values, correlations, data quality
get_feature_importance_toolIdentify top features driving a target variable using ML
train_automl_models_toolAuto-train & compare multiple ML models, returns best performer with R²/accuracy
forecast_timeseries_toolForecast future values using advanced ML models
generate_report_toolFull analysis report including stats, ML insights, and recommendations

All tools accept raw CSV data via a csv_data parameter — no file uploads needed.


⚡ Quick Install

Option 1 — MCPize (Recommended, One-Click)

Install on Claude Desktop Install on Cursor Install on VS Code

  1. Go to mcpize.com and find ProData AI
  2. Click Install for your preferred client
  3. Sign in with your MCPize account (free tier: 50 requests/month)
  4. Start analyzing data!

Option 2 — Claude Code CLI

claude mcp add --transport http prodata-ai https://prodata-ai.mcpize.run \
  --header "API_KEY: your_mcpize_api_key"

Option 3 — Claude Desktop claude_desktop_config.json

{
  "mcpServers": {
    "prodata-ai": {
      "type": "http",
      "url": "https://prodata-ai.mcpize.run",
      "headers": {
        "API_KEY": "your_mcpize_api_key"
      }
    }
  }
}

Get your API key from mcpize.com after signing up.


💬 Example Usage

Once connected, just talk to Claude naturally:

"Analyze this CSV and tell me about data quality"
"Which features most influence sales revenue?"
"Train ML models on this dataset, target column is revenue"
"Forecast the next 30 days of sales"
"Generate a full report including ML insights"

🏗️ Architecture

User (Claude Desktop / Cursor / VS Code)
        ↓  MCP Protocol (HTTP)
MCPize Gateway (auth + billing + routing)
        ↓
Railway Server — server_http.py (FastMCP + Starlette)
        ↓
5 ProData AI Tools (scikit-learn, pandas, statsmodels)

🚀 Self-Hosting / Local Development

Prerequisites

  • Python 3.13+
  • pip

Setup

git clone https://github.com/Varu4/prodata-ai-mcp.git
cd prodata-ai-mcp
pip install -r requirements.txt

Run locally

python server_http.py

Server starts at http://localhost:8080/mcp

Environment Variables

# .env (copy from .env.example)
PORT=8080

⚠️ Never commit real secrets. All sensitive keys go in Railway's dashboard or your local .env file (gitignored).


📦 Tech Stack

LayerTechnology
MCP FrameworkFastMCP + Starlette
RuntimePython 3.13
MLscikit-learn, pandas, numpy
Forecastingstatsmodels
HostingRailway
MarketplaceMCPize (80% revenue share)
CI/CDGitHub Actions → Railway auto-deploy

📊 Performance Benchmarks

Tested on a 120-row retail sales dataset:

ModelR² Score
Gradient Boosting0.9866 ✅ Best
Random Forest0.9741
Linear Regression0.8923

Top Feature: Marketing Spend → Revenue (identified by get_feature_importance_tool)


🗺️ Roadmap

  • analyze_dataset_tool
  • get_feature_importance_tool
  • train_automl_models_tool
  • forecast_timeseries_tool
  • generate_report_tool
  • clean_dataset_tool (coming soon)
  • generate_dashboard_tool — returns self-contained HTML dashboard (coming soon)

🤝 Contributing

Contributions are welcome! Feel free to open issues or submit PRs.

  1. Fork the repo
  2. Create your branch: git checkout -b feature/your-feature
  3. Commit your changes: git commit -m 'Add some feature'
  4. Push: git push origin feature/your-feature
  5. Open a Pull Request

👨‍💻 Author

Varun Walekar — Data Analyst & AI Developer, Bengaluru

GitHub Gumroad Demo


📄 License

MIT License — see LICENSE for details.


If ProData AI saved you time, give it a ⭐ — it helps others find it!

Ver o original no GitHub

Bom para

Relacionados