Forecasting port expenses with ML micro-services

Challenge

Automating procurement of supplies and services when vessels visit ports reduces cost — but required reliable forecasting of port expenses such as tug-boat usage.

Solution

A forecasting algorithm was implemented as a containerised solution, ready for deployment as a micro-service in Maersk's landscape.

• Smart estimation. The model uses vessel and terminal information to estimate how many tug boats will be needed at arrival and departure.
• Instant & reproducible. The automated estimate saves procurement teams time and feeds a reliable, instant prediction straight into the procurement system.

Results

• +5% better classification accuracy
• Micro-service delivered
• 1 month development time