Скачать или смотреть Team 9: Estimated Time of Arrival

Our team has been working on predicting the shipment arrival times and shipment trajectories of different vessels. Our task has been divided into three parts. First, we had to handle 200GB data of vessel locations and converted them into SQL table, and find out the port names with port locations through web scraping. Then we worked on unsupervised trajectory clustering, to make sure we find the trajectories of the meaningful journeys, as well as the starting and ending points. We also solved the problem of unequal interval time series by an innovative solution of fitting a very high degree polynomial and sample it at equal time instances. After the above two steps, our final dataset is 20k trajectories in training and 15k in testing. In our last step, we’ve implemented linear Regression model on predicting the trajectory path. Please watch our recorded video for more details!