BI & Analytics Case Study Enabling Data-Driven Port Service Cost Prediction for International Vessels With 95% Accuracy Port Service costs form a major portion of overall voyage costs. When vessels have access to accurate Port Service cost estimates, it helps in planning dock time, repairs, and frequency of port calls. But very often due to dependencies on vessel size, port tariffs and other dynamic considerations, providers struggle to deliver accurate port service cost prediction. Our client, one of the world’s largest facility services provider, catered to several international carriers, with services offered across the globe in all major ports. They had access to massive amounts of customer data but were not able to use it optimally to predict Port Service cost and create estimates. The difference between estimates and invoice was around a massive 45% in all the international ports! With the right tools and technologies, and by harnessing our Decision Sciences Lab’s AI/ML-based solutions, we enabled the following: Advanced statistical algorithms like Gradient Boosting and XG Boost were tried and tested to achieve 95 % accuracy International ports and 3 major clients were piloted The model was deployed as a FaaS in AWS Cloud to support both Live and Batch predictions Business Benefits: 95% accuracy in prediction Generic ML model catering to all ports Deployed in Cloud – Ease to integrate with other applications The variation was reduced to much below 10% between the estimates and invoices.