A bi-level Random Forest based approach for estimating O-D matrices: Preliminary results from the Belgium National Household Travel Survey

Abstract This paper presents a random forests (RF) based approach to estimate an origin-destination (O-D) matrix on the basis of a travel survey. The trips are predicted on a weekly basis to retain a maximum number of recorded trips for model calibration and validation. The flexibility of the procedure ensures an extension for further disaggregate estimates of O-D matrices. We adopt data stemming from the Belgium National Household Travel Survey as an input for estimating the O-D matrix, in contrast to conventional approaches that exploit traffic counts. Regarding the methodology, preliminary results indicate that the RF approach provides interesting approximations of the O-D traffic flows. The mix of “bagging” and “random subspace” principles included in the RF framework confines the risk of overfitting. Furthermore, the approach is capable of handling large dataset in terms of the number of features and the number of observations.

[1]  S. Bera,et al.  Estimation of origin-destination matrix from traffic counts: the state of the art , 2011 .

[2]  Tomer Toledo,et al.  Estimation of Dynamic Origin–Destination Matrices Using Linear Assignment Matrix Approximations , 2013, IEEE Transactions on Intelligent Transportation Systems.

[3]  Yousef Shafahi,et al.  The treatment of uncertainty in the dynamic origin–destination estimation problem using a fuzzy approach , 2015 .

[4]  Michel Bierlaire,et al.  Bayesian Demand Calibration for Dynamic Traffic Simulations , 2011, Transp. Sci..

[5]  Yali Amit,et al.  Shape Quantization and Recognition with Randomized Trees , 1997, Neural Computation.

[6]  Francesco Viti,et al.  Dynamic Origin–Destination Matrix Estimation on Large-Scale Congested Networks Using a Hierarchical Decomposition Scheme , 2014, J. Intell. Transp. Syst..

[7]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[8]  M. Maher.  INFERENCES ON TRIP MATRICES FROM OBSERVATIONS ON LINK VOLUMES: A BAYESIAN STATISTICAL APPROACH , 1983 .

[9]  Serge P. Hoogendoorn,et al.  Methodology for efficient real time OD demand estimation on large scale networks , 2014 .

[10]  Davy Janssens,et al.  Bayesian inference for transportation origin–destination matrices: the Poisson–inverse Gaussian and other Poisson mixtures , 2015, 2011.06045.

[11]  Davy Janssens,et al.  A Bayesian Approach For Modeling Origin-Destination Matrices , 2011 .

[12]  Fulvio Simonelli,et al.  Limits and perspectives of effective O-D matrix correction using traffic counts , 2009 .

[13]  F. Viti,et al.  Dynamic origin–destination estimation in congested networks: theoretical findings and implications in practice , 2013 .

[14]  Michael G.H. Bell,et al.  The Estimation of an Origin-Destination Matrix from Traffic Counts , 1983 .

[15]  Geert Wets,et al.  Assessing the Quality of Origin–Destination Matrices Derived from Activity Travel Surveys , 2010 .

[16]  E. Cascetta.  Estimation of trip matrices from traffic counts and survey data: A generalized least squares estimator , 1984 .

[17]  Matteucci Matteo,et al.  OD Matrices Network Estimation from Link Counts by Neural Networks , 2013 .

[18]  Moshe E. Ben-Akiva,et al.  Estimation and Prediction of Time-Dependent Origin-Destination Flows with a Stochastic Mapping to Path Flows and Link Flows , 2002, Transp. Sci..

[19]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[20]  Stephen Greaves,et al.  Household travel surveys: Where are we going? , 2007 .