United KingdomTravel and aviation2024 – ongoingHeathrow
Machine learning passenger-flow forecasting
Heathrow cut flight-level forecast error from 30% to 10% after moving passenger-flow forecasting from spreadsheets to machine learning
Problem
Heathrow built its passenger-flow forecasts in spreadsheets. A forecast took about two weeks and two people, and errors were high. Data sat in separate legacy tools, which slowed processing and raised costs.
Approach
Heathrow moved to the Databricks Data + AI Platform on Azure and built a machine learning model for passenger flow. It trains and tracks its models with MLflow. The page does not name the algorithms.
Results
Takeaway
Heathrow replaced a two-person, two-week spreadsheet job with a model. The page does not say how the model is used in daily airport planning.
Categories
- Industry
- Travel and aviation
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Databricks describes how Heathrow replaced spreadsheet-based passenger-flow forecasting, which took two weeks and two people, with a machine learning model on its platform. It reports data processing cut from 80 hours to 4, labor costs down 50%, and flight-level forecast error down from 30% to 10%.
Full details are in the original sources.
