Open Accessbrief discussion on the treatment of spatial correlation in multinomial discrete models
Author(s) -
Francisco J. Bahamonde-Birke
Publication year - 2021
Publication title -
journal of transport and land use
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.05
H-Index - 27
ISSN - 1938-7849
DOI - 10.5198/jtlu.2021.1848
Subject(s) - multinomial distribution , correlation , spatial correlation , discrete choice , computer science , econometrics , compromise , mathematics , sociology , telecommunications , social science , geometry
Spatial dependence plays a key role in all phenomena involving the geographic space, such as the social processes associated with transport and land use. Nevertheless, spatial dependence in multinomial discrete models has not received the same level of attention as have the other kinds of correlations in the discrete modeling literature, mainly due to the complexity of its treatment. This paper aims at offering a brief discussion on the different kinds of spatial correlation affecting multinomial discrete models and the different ways in which spatial correlation has been addressed in the discrete modeling literature. Furthermore, the paper offers a discussion on the advantages and limitations of the different approaches to treat spatial correlation and it also proposes a compromise solution among complexity, computational costs, and realism that can be useful in some specific situations.