Traffic incident duration prediction via a deep learning framework for text description encoding

TL;DR

This paper introduces a deep learning framework that combines traffic flow data and incident descriptions to accurately predict traffic incident durations, significantly outperforming traditional models.

Contribution

It presents a novel fusion deep learning approach integrating incident text encoding and traffic data for improved duration prediction accuracy.

Findings

60% improvement over linear and support vector regression models

7% additional improvement over hybrid deep learning auto-encoded GBDT

Effective in the city of San Francisco with rich incident and traffic data

Abstract

Predicting the traffic incident duration is a hard problem to solve due to the stochastic nature of incident occurrence in space and time, a lack of information at the beginning of a reported traffic disruption, and lack of advanced methods in transport engineering to derive insights from past accidents. This paper proposes a new fusion framework for predicting the incident duration from limited information by using an integration of machine learning with traffic flow/speed and incident description as features, encoded via several Deep Learning methods (ANN autoencoder and character-level LSTM-ANN sentiment classifier). The paper constructs a cross-disciplinary modelling approach in transport and data science. The approach improves the incident duration prediction accuracy over the top-performing ML models applied to baseline incident reports. Results show that our proposed method can improve the accuracy by $60\%$ when compared to standard linear or support vector regression models, and a further $7\%$ improvement with respect to the hybrid deep learning auto-encoded GBDT model which seems to outperform all other models. The application area is the city of San Francisco, rich in both traffic incident logs (Countrywide Traffic Accident Data set) and past historical traffic congestion information (5-minute precision measurements from Caltrans Performance Measurement System).