Description:

This was a data science project and associated web API to predicting cargo vessels' next ports of call and arrival times

Eventually, this tool was able to predicted vessel's next 3 port calls 50% of the time, for any non-oil tanker shipping vessel in the world. In instances when the next port call was predicted correctly, the tool predicted port call times within 10 hours, over 50% of the time.

My role involves a large amount of data engineering, followed by designing and trying algorithms. Finally, I was able to write a RESTful server in Flask.

Some techniques I used include data interpolation, sequence generators, DNN's, Markov chains, and many auto ML ensembles

Meta:

• Start: 2018
• End: 2019
• Location: Berlin, Germany
• Role(s):
• Research Programmer