Airline Route Intelligence Assistant

An LLM-powered aviation analytics assistant that answers natural language questions about airline routes, passenger traffic, load factors, and network trends using real aviation datasets. The system combines a structured aviation dataset with LLM tool-calling to generate grounded analytical insights instead of hallucinated responses.

Overview

Airline network planning and route analysis typically require querying complex datasets and interpreting aviation metrics such as passenger volume, load factor, and departures. This project demonstrates how Large Language Models can act as an intelligent interface for aviation analytics by translating natural language questions into structured data queries and returning analytical summaries.

Example queries supported by the system:

1. Compare PHX–LAX and PHX–SFO
2. Which routes have the lowest load factor?
3. Show the busiest routes in the network
4. Summarize traffic trends for a route
5. Analyze route JFK–LAX The assistant uses tool-calling to query structured data and return responses grounded in real metrics.

Features

Natural language query interface LLM tool-calling for structured aviation analytics Route comparison with visual charts Follow-up question support Interactive web interface built with Streamlit Grounded responses using dataset queries (no hallucinated answers)

System Architecture

The application follows a modular architecture separating UI, LLM orchestration, and analytics tools. User Query │ ▼ Streamlit Web Interface │ ▼ LLM Service (OpenAI) │ ├── Query Router │ ├── Tool Calling │ ▼ Analytics Engine │ ├── Route Analysis ├── Route Comparison ├── Network Summary │ ▼ Dataset (Airline Traffic Data)

The LLM decides which analytical tool to call based on the user's query and then synthesizes a response using the returned metrics.

Project Structure

airline-route-intelligence-assistant

app/ ├── services/ │ └── llm_service.py │ ├── tools/ │ └── analytics_tools.py │ └── query_router.py

data/ └── airline_routes_dataset.csv

vectorstore/

tests/ ├── test_llm_service_manual.py └── test_query_router_manual.py

assets/ ├── logo.png └── banner.png

streamlit_app.py

requirements.txt README.md .env

Example Query

User input:

"Compare PHX-LAX and PHX-SFO"

Assistant response:

PHX–LAX carries significantly higher passenger traffic compared to PHX–SFO.
The PHX–LAX route also demonstrates stronger load factors and more frequent departures. PHX–SFO shows lower passenger density per departure, indicating weaker route demand relative to PHX–LAX.

Installation

Clone the repository:

git clone https://github.com/yourusername/airline-route-intelligence-assistant.git cd airline-route-intelligence-assistant

Install dependencies: pip install -r requirements.txt Create a .env file and add your OpenAI API key: OPENAI_API_KEY=your_api_key_here

Running the Application

Start the Streamlit interface: streamlit run streamlit_app.py

Technologies Used

Python Streamlit OpenAI API Pandas LLM Tool Calling Data Analytics

Example Use Cases

Airline route demand comparison Network traffic analysis Passenger load factor investigation Data-driven route insights using natural language

Future Improvements

RAG pipeline for aviation reports Live aviation datasets Geospatial route visualization Airline profitability estimation Multi-agent aviation analysis

Author

Nibras Dsouza Master's in Software Engineering Arizona State University

To view the demo assets-> screenshots.

Why this project matters

This project demonstrates how LLMs can be integrated with structured analytics systems to transform natural language queries into grounded insights for real-world domains such as aviation network planning.

Architecture

User Query ↓ Streamlit Interface ↓ LLM Tool Router ↓ Route Analytics Engine ↓ Structured Response + Visualization