Part 1: Destination Model Whitepaper
Released in 2023, part one of this whitepaper series uncovers the multifaceted challenge of predicting a tanker’s ultimate destination and the critical need for an adequate solution.
It also introduced the foundational concepts of the Vortexa machine learning system, designed to leverage vast datasets to accurately predict tanker loading and discharge locations.
A volatile market demands greater predictive ability
In this whitepaper, we unveil how Vortexa’s innovative use of transformer models improves the accuracy of destination predictions by tapping into the deeper contexts influencing vessel behaviour. By treating a vessel’s historical activity as a sequence of events akin to words in a sentence, we leverage transformers’ unique ability to capture contextual relationships within time series data.
With this innovative approach, we have not only taken a significant step in the evolution of maritime analytics, but also achieved a remarkable 10% relative improvement in accuracy compared to our previous models.
Part 2: Destination Sequence Model Whitepaper
In part two of our whitepaper series on Vortexa modelling, we are
delighted to introduce a unique machine-learning model for vessel destination predictions.
This model leverages the power of transformers to unlock and decode
new patterns hidden within vast streams of maritime data.