The classic transport model consists of four sub-models: trip generation, trip distribution, modal split and traffic assignment. The current article in particular focuses on trip distribution. In this paper a statistical approach is introduced to deal with spatial autocorrelation among errors in a regression model based on commuter data. A gravity model is applied to estimate traffic flows by using regression techniques. Unlike time-series in cross-section analysis the measures of autocorrelation must take into consideration different contiguities among units. In a first step - illustrated by a simple cross-sectional example - a theoretical framework is developed, allowing the estimation of spatial weight matrices adapted from a Moran coefficient as a measure of spatial correlation. These weight matrices establish a basis for estimating a spatial autocorrelation coefficient and, moreover, reflect the various effects of different neighbourhood matrices. Finally, the same approach is applied by using commuter flow data. Here, contrary to simple cross-sectional analysis the contiguity matrix must take into account vectors. It is shown, that in travel flow data as well as in cross-sectional data spatial autocorrelation can be found. Accounting for this phenomenon by an appropriate spatially autocorrelated structure of the error terms leads to significant improvements of the transport flow model.
Neuere oekonometrische Methoden in der Verkehrswissenschaft
Advanced econometric methods for traffic science
Zeitschrift für Verkehrswissenschaft ; 75 , 1 ; 25-49
2004
25 Seiten
Article (Journal)
German
Transport , Modell , Detail , Lösungsansatz , Autokorrelation , Verkehrsablauf , Fachwerk , Gewicht , Korrelation , Matrix , Rechnungswesen , Term , Abstand , Berechnungsmethode , Beweglichkeit , Fahrzeug , Forschungsprogramm , Korrelationsanalyse , Optimierung , Wirtschaftlichkeit , Programm , Rechenmodell , Soziologie , Sozialwissenschaft , Verkehrsmittel , Wechselbeziehung , Wissenschaft , Wichtungsfunktion , geometrische Größe , Statistik , mathematische Statistik , Forschung , Angebot
Neuere ökonometrische Methoden in der Verkehrswissenschaft
Online Contents | 2004
|Neuere ökonometrische Methoden in der Verkehrswissenschaft
IuD Bahn | 2004
|SLUB | 1968
|Beiträge zur Verkehrswissenschaft
SLUB | 1978
|ZEITSCHRIFT FÜR VERKEHRSWISSENSCHAFT
Online Contents | 2006