DOI: 10.1068/A040205
关键词: Spatial interaction 、 Entropy (information theory) 、 Artificial intelligence 、 Iterative and incremental development 、 Rate of convergence 、 Fibonacci search technique 、 Gravity model of trade 、 Applied mathematics 、 Simplex algorithm 、 Multivariate statistics 、 Computer science
摘要: This paper presents a methodology for deriving best statistics the calibration of spatial interaction models, and several procedures finding parameter values are described. The family models due to Wilson is first outlined, then some existing methods briefly reviewed. A procedure based on principle maximum-likelihood developed from work Hyman Evans, illustrated using example retail gravity model. Five solving equations outlined: simple first-order iterative process, Newton—Raphson method variables, multivariate Fibonacci search, search Simplex method, quadratic convergence, all tested compared. It appears that most efficient, this further in disaggregated residential location models.