Adaptive Kernel Density Estimation for Analysing Spatio-Temporal Point Patterns – An Application to Food Retailing Markets

Description

This research reviews the method of adaptive kernel density estimation for the modelling and exploration of spatial point patterns with a particular reference to spatio-temporal variations in food retailing concentration and local micro market structures. The case study region Upper Austria comprises sub-areas in which grocery outlets (i.e. the events) are more densely packed than others. Thus, adaptive local adjustment of bandwidth seems to be appropriate for balancing the reliability of estimates against the degree of spatial detail that is retained, given the observed pattern of event locations. In order to illustrate this technique exemplarily, the changes in spatial concentration of grocery outlets of two groups of retailers are analysed for two dates. One group comprises the two market leaders (RS), which had a joint market share in 2004 of about 71 percent in terms of total revenue, and of 45 percent in terms of outlets, respectively. The other group consists of the other market participants (OMP). The discussion will centre on the evaluation of adaptive dual density kernel estimates compared to results obtained in a previous study which used dual kernel density estimation with fixed bandwidth.

Period	27 Aug 2008 → 31 Aug 2008
Event title	ERSA 2008: 48th Congress of the European Regional Science Association
Event type	Unknown
Degree of Recognition	International