Artificial Neural Networks Applied to Taxi Destination Prediction

TL;DR

This paper presents a neural network-based solution for taxi destination prediction, achieving first place in a competitive challenge by effectively modeling variable-length GPS sequences and associated metadata.

Contribution

It introduces an automated neural network approach using various architectures for sequence-to-fixed output prediction, outperforming prior methods in taxi destination forecasting.

Findings

Achieved first place among 381 teams in the challenge.

Successfully modeled variable-length GPS sequences with neural networks.

Demonstrated adaptability to other sequence-to-fixed output prediction tasks.

Abstract

We describe our first-place solution to the ECML/PKDD discovery challenge on taxi destination prediction. The task consisted in predicting the destination of a taxi based on the beginning of its trajectory, represented as a variable-length sequence of GPS points, and diverse associated meta-information, such as the departure time, the driver id and client information. Contrary to most published competitor approaches, we used an almost fully automated approach based on neural networks and we ranked first out of 381 teams. The architectures we tried use multi-layer perceptrons, bidirectional recurrent neural networks and models inspired from recently introduced memory networks. Our approach could easily be adapted to other applications in which the goal is to predict a fixed-length output from a variable-length sequence.