Shipping Vessel Prediction

Description

This was a data science project and associated web API for predicting cargo vessels' next ports of call and arrival times. It achieved the following:

  • Predicted vessels' next three port calls with 50% accuracy.
  • Predicted port call times within a 10-hour margin in over 50% of correct predictions.

My role involved: - Data engineering. - Designing and testing algorithms. - Writing a RESTful API server in Flask.

Techniques used included data interpolation, sequence generators, deep neural networks (DNNs), Markov chains, and automated ML ensembles.

Meta

  • Start Date: 2018
  • End Date: 2019
  • Location: Berlin, Germany
  • Role: Research Programmer

Visit Latentine