Feature Engineering for Supply Analysis in Ocean Transportation

In recent years, much effort has been put into the application of ship tracking data for the purpose of shipping market modeling. Such high-frequency spatial data enables real time monitoring of the whereabouts of the global fleet and the activity along the networks on which it trades. However, for spatial data to be of use for price prediction it needs to be enhanced with commercial data, only part of which is publicly available. In this paper we expand on the inherent challenges in deriving the true demand and supply balance in the freight market and show examples of advanced feature engineering using merged spatial and commercial data.