A gradient boosting method to improve travel time prediction

By Yanru Zhang, Ali Haghani. 2015

782 citations recorded by the source corpus.

Tree based ensemble methods have reached a celebrity status in prediction field. By combining simple regression trees with ‘poor’ performance, they usually produce high prediction accuracy. In contrast to other machine learning methods that have been treated as black-boxes, tree based ensemble methods provide interpretable results, while requiring little data preprocessing, are able to handle different types of predictor variables, and can fit complex nonlinear relationship. These properties make the tree based ensemble methods good candidates for solving travel time prediction problems. However, applications of tree-based ensemble algorithms in traffic prediction area are limited. In this paper, we employ a gradient boosting regression tree method (GBM) to analyze and model freeway travel time to improve the prediction accuracy and model interpretability. The gradient boosting tree method strategically combines additional trees by correcting mistakes made by its previous base models, therefore, potentially improves prediction accuracy. Different parameters’ effect on model performance and correlations of input–output variables are discussed in details by using travel time data provided by INRIX along two freeway sections in Maryland. The proposed method is, then, compared with another popular ensemble method and a bench mark model. Study results indicate that the GBM model has its considerable advantages in freeway travel time prediction.

Subfields: Building and Construction, Control and Systems Engineering, Transportation

View Paper: https://www.semanticscholar.org/paper/8142fad2964a23636818ca33c98aa063e270f3e5

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Building and Construction

In Control and Systems Engineering

In Transportation

Citations in the corpus

Travel time prediction with LSTM neural network (2016, methodology): “The gradient boosting regression tree method [4] combines simple regression trees to produce high accuracy travel time prediction with previous multiple time periods travel times and related variables as model inputs.”
CatBoost: unbiased boosting with categorical features (2017, methodology): “For a number of years, it has remained the primary method for learning problems with heterogeneous features, noisy data, and complex dependencies: web search, recommendation systems, weather forecasting, and many others [5, 19, 21, 23].”
CatBoost: gradient boosting with categorical features support (2018, methodology): “For a number of years, it has remained the primary method for learning problems with heterogeneous features, noisy data, and complex dependencies: web search, recommendation systems, weather forecasting, and many others [2, 15, 17, 18].”
A Survey of Ensemble Learning: Concepts, Algorithms, Applications, and Prospects (2022, positive): “However, the intrinsic approach to regularize the GBDT is by using shrinkage to limit every gradient decent , where v is normally assigned a value of 0.1 [82].”
A boosted decision tree approach using Bayesian hyper-parameter optimization for credit scoring (2017, background): “Recent studies show that boosting technique is a promising approach in the domain of insurance loss modeling (Guelman, 2012), travel time prediction (Yanru Zhang & Haghani, 2015), and chemistry (Brillante et al., 2015).”
LSTM network: a deep learning approach for short-term traffic forecast (2017, background): “…this problem, researchers also paid much attention to non-parametric approaches in the traffic flow forecasting field, such as non-parametric regression [17], neural network prediction [18], support vector machine (SVM) [19], Kaman filtering [20, 21] and the combination of these algorithms [22–27].”
Prediction and Sensitivity Analysis of Bubble Dissolution Time in 3D Selective Laser Sintering Using Ensemble Decision Trees (2019, methodology): “EDT Bagged is a hybrid model which is a combination of bagging algorithm and decision trees [44].”
A review of ensemble learning and data augmentation models for class imbalanced problems: combination, implementation and evaluation (2023, background): “…credit risk assessment [125], mobility prediction [126], anti-money laundering [127], optimisation of transduction [128], strength prediction [129], travel time prediction [130], modelling energy consumption [131], geography [132, 133], medical diagnosis [134, 135, 136], image recognition [137].”
Road Traffic Forecasting: Recent Advances and New Challenges (2018, background): “Exogenous factors continue to be barely addressed in recent research, and calendar information is the most considered one [9], [32], [44], [46], [55], [61], [69], [76], [87], [88], [93], [104], [107].”