Sensitivity Analysis and Optimisation of a Land Use and Transport Integrated Model

Gilquin, Laurent; Capelle, Thomas; Arnaud, Elise; Prieur, Clémentine

Journal de la société française de statistique, Special Issue : Computer Experiments, Uncertainty and Sensitivity Analysis, Tome 158 (2017) no. 1, pp. 90-110.

Les modèles « transport-urbanisme » sont devenus une norme pour représenter les interactions entre l’usage des sols et le transport de marchandises et d’individus. Ces modèles sont principalement utilisés dans le cadre d’évaluations de politiques d’urbanisme et de scénarios de développement urbain. Le calage des modèles « transport-urbanisme » est une tâche difficile qui nécessite l’estimation d’un nombre important de paramètres. Dans ce papier, nous considérons le calage du modèle en libre accès Tranus. Une estimation classique des paramètres de Tranus repose à la fois sur des techniques ad hoc d’économétrie et sur des procédures de type essais-erreurs coûteuses en temps. Dans ce papier, nous proposons un calage en deux étapes comprenant une phase d’analyse de sensibilité globale et une phase d’optimisation itérative. La méthode d’analyse de sensibilité présentée ici est basée sur la méthode répliquée, estimant des indices de Sobol’, et généralisée au cas de sorties multidimensionnelles. La phase d’optimisation est une procédure itérative combinant deux approches : une stochastique et une analytique. La méthode de calage est appliquée à la zone d’étude dans l’Etat du Mississippi. Par comparaison avec une précédente méthode de calage ad hoc, notre approche aboutit à une amélioration significative des facteurs d’ajustement de Tranus avec un temps de calage considérablement réduit.

Land Use and Transportation Integrated (LUTI) models have become a norm for representing the interactions between land use and the transportation of goods and people in a territory. Through the use of these models, urban planning policies and development scenarios can be evaluated. The calibration of LUTI models is a heavy task, involving gathering of massive amounts of data and the estimation of an important number of parameters. In this paper, the calibration of the open-source LUTI model Tranus is considered. Classical calibrations of Tranus rely on ad hoc econometric techniques and time-consuming trial and error procedures.Here, a two-step calibration that comprises global sensitivity analysis and optimisation is proposed. The sensitivity analysis presented herein is based on the replication method for the estimation of Sobol’ indices and generalised to take into account multivariate outputs. The optimisation step is an iterative process combining stochastic and deterministic procedures. The proposed calibration procedure is applied to a study area in the State of Mississippi. Compared to a previous ad hoc procedure, this new approach results in a significant improvement of the adjustment factors of Tranus while reducing drastically the calibration time.

[1] Abraham, J. E.; Hunt, J. D. Parameter Estimation Strategies for Large-Scale Urban Models, Transportation Research Record: Journal of the Transportation Research Board, Volume 1722 (2000) no. 1, pp. 9-16

[2] Ciuffo, B.; Azevedo, C. L. A Sensitivity-Analysis-Based Approach for the Calibration of Traffic Simulation Models, IEEE Trans. Intell. Transp. Syst., Volume 15 (2014) no. 3, pp. 1298-1309

[3] Caflisch, R. E.; Morokoff, W. J.; Owen, A. B. Valuation of mortgage backed securities using Brownian bridges to reduce effective dimension, J. Comput. Finance, Volume 1 (1997) no. 1, pp. 27-46

[4] Capelle, T.; Sturm, P.; Vidard, A.; Morton, B. J. Formulating LUTI Calibration as an Optimisation Problem: Example of Tranus Shadow Price Estimation, Procedia Engineering, Volume 115 (2015), pp. 12-20

[5] Daamen, W.; Buisson, C.; Hoogendoorn, S. P. Traffic Simulation and Data: Validation Methods and Applications, Taylor & Francis Group, 2014

[6] de la Barra, T. Mathematical description of TRANUS (1999) http://www.tranus.com/tranus-english (Technical report)

[7] Dutta, P.; Saujot, M.; Arnaud, E.; Lefevre, B.; Prados, E. Uncertainty Propagation and Sensitivity Analysis During Calibration of TRANUS, an Integrated Land Use and Transport Model, International Conference on Urban, Regional Planning and Transportation (2012)

[8] Ge, Q.; Ciuffo, B.; Menendez, M. An Exploratory Study of Two Efficient Approaches for the Sensitivity Analysis of Computationally Expensive Traffic Simulation Models, IEEE Trans. Intell. Transp. Syst., Volume 15 (2014) no. 3, pp. 1288-1297

[9] Gamboa, F.; Janon, A.; Klein, T.; Lagnoux, A. Sensitivity analysis for multidimensional and functional outputs, Electron. J. Statist., Volume 8 (2014) no. 1, pp. 575-603 | MR | Zbl

[10] Ge, Q.; Menendez, M. An efficient sensitivity analysis approach for computationally expensive microscopic traffic simulation models, International Journal of Transportation, Volume 2 (2014) no. 2, pp. 49-64

[11] Hoeffding, W.F. A class of statistics with asymptotically normal distributions, Annals of Mathematical Statistics, Volume 19 (1948) no. 3, pp. 293-325 | MR | Zbl

[12] Janon, A.; Klein, T.; Lagnoux, A.; Nodet, M.; Prieur, C. Asymptotic normality and efficiency of two Sobol’ index estimators, ESAIM Probab. Stat., Volume 18 (2014), pp. 342-364 | MR | Zbl

[13] Jones, D. R.; Schonlau, M.; Welch, W. J. Efficient global optimization of expensive black-box functions, J. Global Optim., Volume 13 (1998), pp. 455-492 | MR | Zbl

[14] Leontief, W. W. The Structure of the American Economy 1919-1939, New York: Oxford University Press, 1941

[15] Leontief, W. W.; Strout, A. Multi-Regional Input-Output Analysis, Structural Interdependence and Economic Development, London: Mcmillan, 1963

[16] McFadden, D. Conditional logit analysis of qualitative choice behaviour, Frontiers in Econometrics - Academic Press New York (1973), pp. 105-142

[17] McKay, M. D. Evaluating prediction uncertainty (1995) (Technical report)

[18] Mara, T. A.; Joseph, O. Rakoto Comparison of some efficient methods to evaluate the main effect of computer model factors, Journal of Statistical Computation and Simulation, Volume 78 (2008) no. 2, pp. 167-178 | Zbl

[19] Morris, M. D.; Moore, L.; McKay, M. D. Sampling plans based on balanced incomplete block designs for evaluating the importance of computer model inputs, J. Statist. Plann. Inference, Volume 136 (2006) no. 9, pp. 3203-3220 | Zbl

[20] Morris, M. D.; Moore, L.; McKay, M. D. Orthogonal Arrays in the Sensitivity Analysis of Computer Models, Technometrics, Volume 50 (2008), pp. 205-215

[21] Monod, H.; Naud, C.; Makowski, D. 3, Uncertainty and sensitivity analysis for crop models, Elsevier (2006), pp. 55-100

[22] Morris, M. D. Factorial Sampling Plans for Preliminary Computational Experiments, Technometrics, Volume 33 (1991) no. 2, pp. 161-174

[23] McFadden, D.; Train, K. Mixed MNL Models for Discrete Response, Journal of Applied Econometrics, Volume 15 (2000), pp. 447-470

[24] Mockus, J.; Tiesis, V.; Zilinskas, A. The application of Bayesian methods for seeking the extremum, Towards Global Optimisation, Vol. 2 (Dixon, L. C. W.; Szego, G. P., eds.), North-Holand, 1978, pp. 117-129 | Zbl

[25] Ortúzar, J. D.; Willumsen, L. G. Modelling Transport, Wiley, 2011

[26] Roustant, O.; Ginsbourger, D.; Deville, Y. DiceKriging, DiceOptim: Two R Packages for the Analysis of Computer Experiments by Kriging-Based Metamodeling and Optimization, J. Stat. Softw, Volume 51 (2012) no. 3

[27] Saltelli, A. Making best use of model evaluations to compute sensitivity indices, Comput. Phys. Commun., Volume 145 (2002) no. 2, pp. 280-297 | Zbl

[28] Sobol’, I. M. Sensitivity indices for nonlinear mathematical models, Mathematical Modeling and Computational Experiment, Volume 1 (1993), pp. 407-414 | Zbl

[29] Tissot, J. Y.; Prieur, C. A Randomized Orthogonal Array-based procedure for the estimation of first- and second-order Sobol’ indices, J. Statist. Comput. Simulation, Volume 85 (2015), pp. 1358-1381 | Zbl

[30] Wilson, A. G. Optimization in locational and transport analysis, Wiley & Sons, New York, 1981 | Zbl