On the Data-Driven Prediction of Arrival Times for Freight Trains on U.S. Railroads

William Barbour, Chinmaya Samal, Shankara Kuppa, Abhishek Dubey, Daniel B. Work

21st International Conference on Intelligent Transportation Systems, ITSC 2018, Maui, HI, USA, November 4-7, 2018 2018

PDF DOI

Why This Matters

Accurate prediction of freight train arrival times is critical for railroad operations, enabling better scheduling, reducing costs, and improving efficiency. Traditional analytical approaches fail to capture the complex relationships between numerous operational factors and actual train delays. This work is innovative because it leverages large historical datasets and modern machine learning to achieve significantly better prediction accuracy than statistical baselines, enabling data-driven operational planning.

What We Did

This paper develops machine learning models to predict freight train arrival times using historical operational data from US railroads. The work compares multiple approaches including support vector regression, random forest, and deep neural networks trained on extensive historical train data covering physical characteristics, crew information, network state, and occupancy metrics. The models account for the high variability in train operations caused by congestion, infrastructure constraints, and scheduling factors.

Key Results

The deep neural network model achieved 26% error reduction compared to statistical baseline predictors, with effective prediction of discrete arrival times along train routes. Support vector regression and random forest models also demonstrated strong performance, significantly outperforming baseline approaches on a five-year historical dataset of over 170,000 labeled examples. The results enable railroad operators to make more informed decisions about crew scheduling and train coordination.

Full Abstract

Cite This Paper

@inproceedings{Barbour2018,
  author = {Barbour, William and Samal, Chinmaya and Kuppa, Shankara and Dubey, Abhishek and Work, Daniel B.},
  booktitle = {21st International Conference on Intelligent Transportation Systems, {ITSC} 2018, Maui, HI, USA, November 4-7, 2018},
  title = {On the Data-Driven Prediction of Arrival Times for Freight Trains on {U.S.} Railroads},
  year = {2018},
  pages = {2289--2296},
  abstract = {The high capacity utilization and the pre-dominantly single-track network topology of freight railroads in the United States causes large variability and unpredictability of train arrival times. Predicting accurate estimated times of arrival (ETAs) is an important step for railroads to increase efficiency and automation, reduce costs, and enhance customer service. We propose using machine learning algorithms trained on historical railroad operational data to generate ETAs in real time. The machine learning framework is able to utilize the many data points produced by individual trains traversing a network track segment and generate periodic ETA predictions with a single model. In this work we compare the predictive performance of linear and non-linear support vector regression, random forest regression, and deep neural network models, tested on a section of the railroad in Tennessee, USA using over two years of historical data. Support vector regression and deep neural network models show similar results with maximum ETA error reduction of 26\% over a statistical baseline predictor. The random forest models show over 60\% error reduction compared to baseline at some points and average error reduction of 42\%.},
  bibsource = {dblp computer science bibliography, https://dblp.org},
  biburl = {https://dblp.org/rec/bib/conf/itsc/BarbourSKDW18},
  category = {selectiveconference},
  contribution = {minor},
  doi = {10.1109/ITSC.2018.8569406},
  file = {:Barbour2018-On_the_Data-Driven_Prediction_of_Arrival_Times_for_Freight_Trains_on_U.S._Railroads.pdf:PDF},
  keywords = {machine learning, train scheduling, arrival time prediction, transportation systems, deep learning},
  project = {smart-transit,cps-reliability,smart-cities},
  tag = {transit},
  timestamp = {Wed, 16 Oct 2019 14:14:57 +0200},
  url = {https://doi.org/10.1109/ITSC.2018.8569406}
}
Quick Info
Year 2018
DOI 10.1109/ITSC.2018.8569406
Keywords
machine learning train scheduling arrival time prediction transportation systems deep learning
Research Areas
transit ML for CPS
Search Tags

Data, Driven, Prediction, Arrival, Times, Freight, Trains, U.S., Railroads, machine learning, train scheduling, arrival time prediction, transportation systems, deep learning, transit, ML for CPS, 2018, Barbour, Samal, Kuppa, Dubey, Work