ErikWrwg/meridian-air-decision-tool

Meridian Air — Climate-Sensitive Demand Decision Tool

An interactive decision tool for a fictional European low-cost carrier balancing revenue against climate commitments.

What this is

A consulting deliverable built for the ESCP "AI for Big Data Management" course project. Meridian Air is a fictional client; the tool answers a real question: can a short-haul European airline forecast demand and price climate-sensitive routes more intelligently than traditional methods?

What's inside

• —Real flight data: 9 years of daily traffic across 20 European airports from Eurocontrol
• —Real climate sentiment: 1,427 Guardian climate-aviation articles scored with VADER
• —Synthetic Meridian bookings: 43,820 daily route-level booking records with calibrated sentiment-driven dynamics
• —Random Forest classifier: identifies climate-sensitive routes from behavioral signals (95% CV accuracy)
• —ARIMA baseline: classical time-series forecast (24% mean test MAPE)
• —LSTM multi-input: deep learning forecast using bookings + sentiment (7% mean test MAPE — 17pp improvement)
• —LLM recommendations: route-level strategic advice via Hugging Face Inference API

How to use

1. 1.Pick a route from the sidebar.
2. 2.Adjust the climate concern scenario slider.
3. 3.Browse the four tabs: demand history, ARIMA-vs-LSTM forecast comparison, classification details, and a generated strategic recommendation.

The story

Meridian's CFO and Chief Sustainability Officer disagree about how to price short-haul routes with viable rail alternatives. This tool gives them a data-driven middle path: route-level climate sensitivity scores, demand forecasts that incorporate climate news cycles, and concrete pricing/offset/marketing recommendations per route.

Project context

ESCP Business School — AI for Big Data Management — Group Project (16% of final grade)