Excel Analysis Agent

A Django-based AI-powered data analysis application that lets users upload Excel/CSV files and ask natural language questions about their data. Built with local LLMs via Ollama, LangChain agents, and vector search.

Features

• File Upload — Upload Excel (.xlsx, .xls) and CSV files with configurable null-value handling strategies (forward fill, backward fill, mean fill, zero fill, etc.)
• Natural Language Chat — Ask questions about your data in plain English; an AI agent translates them into SQL queries, Python computations, or semantic lookups
• Visualization — Auto-generated charts and plots via Matplotlib/Seaborn, plus a dedicated data exploration page with filtering and pagination
• ReAct Agent — Multi-tool reasoning agent with SQL, Python execution, and vector search capabilities
• Streaming Responses — Real-time Server-Sent Events for agent reasoning traces
• Vector Search — Semantic search over your data using ChromaDB embeddings
• Configurable Models — Choose Ollama models for QA, chat, and embeddings via the settings page

Tech Stack

• Backend: Django, LangChain, Pandas, NumPy
• LLM: Ollama (local models, default: qwen3.5:9b)
• Vector Store: ChromaDB with nomic-embed-text embeddings
• Database: SQLite
• Visualization: Matplotlib, Seaborn

Prerequisites

• Python 3.10+
• Ollama installed and running locally

Setup

1. Clone the repository

   git clone <repo-url>
   cd excel_analysis

2. Create a virtual environment and install dependencies

   python -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt

3. Pull the required Ollama models

   ollama pull qwen3.5:9b
   ollama pull nomic-embed-text

4. Run migrations

   python manage.py migrate

5. Start the server

   python manage.py runserver

6. Open http://127.0.0.1:8000/ in your browser.

Environment Variables

Variable	Default	Description
OLLAMA_BASE_URL	http://127.0.0.1:11434	Ollama server URL
OLLAMA_QA_MODEL	qwen3.5:9b	Model for SQL/data queries
OLLAMA_CHAT_MODEL	qwen3.5:9b	Model for conversational responses
OLLAMA_EMBED_MODEL	nomic-embed-text	Embedding model for vector search

These can also be configured from the in-app Settings page.

Usage

1. Upload — Go to /upload and upload an Excel or CSV file. Choose a null-handling strategy if your data has missing values.
2. Chat — Navigate to /chat, create a session, and ask questions about your data (e.g., "What is the average unemployment rate by region?").
3. Visualize — Use /visualization to filter, sort, and paginate through your data.
4. Settings — Configure Ollama models and connection at /settings.

Project Structure

excel_analysis/
├── ai_analysis/            # Main Django app
│   ├── services/           # Business logic
│   │   ├── agent.py        # ReAct agent engine
│   │   ├── data_ingestion.py   # File upload & processing
│   │   ├── data_query.py       # Filtering & visualization
│   │   ├── prompting.py        # Dynamic prompt construction
│   │   ├── app_settings.py     # Ollama config management
│   │   └── runtime.py          # In-memory state management
│   ├── models.py           # Database models
│   ├── views.py            # API endpoints
│   ├── templates/          # HTML templates
│   └── tests/              # Test suite
├── excel_agent/            # Django project settings
├── data/                   # Vector store index
├── media/                  # Uploaded files & generated plots
├── requirements.txt
└── manage.py

Running Tests

pytest

License

This project is for personal/educational use.