Deep Neural Networks for Predicting Vehicle Travel Times

This paper focuses on prediction if vehicle travel time. An established open dataset of taxi trips in New York City is used. We first perform statistical analysis on the data in order to determine the informative features that can be used for the problem at hand. Successive to detailed analysis of the data and features, we develop a deep neural network for travel time prediction. We show that our model performs with high accuracy, and outperforms a number of baseline techniques.