Data Analysis Agent

An interactive, agentic data analysis application that leverages advanced LLM reasoning to help users explore, visualize, and understand their data using NVIDIA Llama-3.1-Nemotron-Ultra-253B-v1.

Overview

This repository contains a Streamlit application that demonstrates a complete workflow for data analysis:

1. Data Upload: Upload CSV files for analysis
2. Natural Language Queries: Ask questions about your data in plain English
3. Automated Visualization: Generate relevant plots and charts
4. Transparent Reasoning: Get detailed explanations of the analysis process

The implementation leverages the powerful Llama-3.1-Nemotron-Ultra-253B-v1 model through NVIDIA's API, enabling sophisticated data analysis and reasoning.

Learn more about the model here.

Features

• Agentic Architecture: Modular agents for data insight, code generation, execution, and reasoning
• Natural Language Queries: Ask questions about your data—no coding required
• Automated Visualization: Instantly generate and display relevant plots
• Transparent Reasoning: Get clear, LLM-generated explanations for every result
• Powered by NVIDIA Llama-3.1-Nemotron-Ultra-253B-v1: State-of-the-art reasoning and interpretability

Requirements

• Python 3.10+
• Streamlit
• NVIDIA API Key
• Required Python packages:
  • pandas
  • matplotlib
  • streamlit
  • requests

Installation

1. Clone this repository

2. Install dependencies:

   pip install -r requirements.txt

3. Set up your NVIDIA API key:

   • Sign up or log in at NVIDIA Build
   • Generate an API key
   • Set the API key in your environment:

     export NVIDIA_API_KEY=your_nvidia_api_key_here

   • Or add it to your .env file if you use one

Usage

1. Run the Streamlit app:

   streamlit run data_analysis.py

2. Use the application:

   • Upload a CSV file (e.g., the Titanic dataset)
   • Ask questions in natural language
   • View results, visualizations, and detailed reasoning

Model Details

The Llama-3.1-Nemotron-Ultra-253B-v1 model used in this project has the following specifications:

• Parameters: 253B
• Features: Advanced reasoning capabilities
• Use Cases: Complex data analysis, multi-agent systems
• Enterprise Ready: Optimized for production deployment