AI/ML-ENABLED AEROSPACE SUPPLY CHAIN MANAGEMENT
A leading American airline faced a hurdle with customer satisfaction regarding their in-flight menu options. The airline struggled to anticipate demand for certain meals, resulting in a surplus of some items and a shortage of others. However, Valorem Reply, a Microsoft Designated Azure Data & AI Solutions Partner, had previous experience working with the airline and had valuable insight into the issue, along with recommendations to address the allocation problem. Within a span of just six weeks, we deployed a Demand Forecasting Accelerator solution that enables the airline to predict catering demand for each flight. Our machine learning model predicts catering demand by item and by flight, providing catering vendors with required menu items, food quantities, etc. in advance. This innovative solution is based on a model produced in Azure ML, which is connected to a new table in Azure SQL DB and runs on the airline's Azure tenant. This provides the airline with access to both the model and the API. With our innovative solution, the airline can now accurately predict customer demand for meals and adjust their stock if there are delays or unexpected changes to flights, facilitating a better customer and employee experience. Consequently, the airline has seen a notable increase in revenue of approximately $2M - $5M per year and a reduction in food waste of $50K - $80K a month.
We are extremely proud of the impact our data science solution has had on addressing a long-standing challenge faced by this airline. By accurately predicting catering demand for each flight, we have not only improved customer satisfaction but also enhanced the brand of the airline by providing consistent service and meeting customer expectations. This achievement has benefited customers, agents, and all areas of the airline, multiplying the positive impact.
Scott Davis, Account Technology Strategist, Valorem Reply
Challenges and Opportunities
• Need to maximize catering revenue and minimize waste.
• Customer satisfaction issues due to menu option availability.
• Improve catering operations and in-flight efficiencies.
• Azure Machine Learning.
• Databricks Engineering Platform.
• Build/refine iterative predictive model.
• Additional variables included in model to include routes, time-of-day, seat configuration, historic load factors.
• Increased demand prediction accuracy by about 90%.
• Reduction in food waste of $50K - $80K a month.
• Increase in revenue of approximately $2M - $5M per year.
• Extensible model available for all menu items.