Statistical machine learning for urban transportation system

In general, the goal of project is to investigate the train travel data and figure out the main factors affecting train travel time. Moreover, we will use machine learning algorithms to predict their arrival time to stations and forecast when delays will happen. Specifically, to figure out what factors are affecting train travel times, we will investigate several possibilities according to prior empirical knowledge. Among them, useful factors will be chosen from the data exploration process and undergone statistical significance tests. Until then, we represent the entire railway map as a graph and build a convolutional neural network based that. The convolutional neural network is a machine learning model can take into account of all affecting factors and generate accurate train arrival time as output. The overall solution can be generalized to any similar railway systems and thus be integrated as a product for the partner organization.

Faculty Supervisor:

Linglong Kong

Student:

Yue Wang

Partner:

NTwist

Discipline:

Mathematics

Sector:

University:

University of Alberta

Program: