Spatial self-organization in Santiago: methods and applications.

Abstract

Assembling spatial units into meaningful clusters is a challenging task, as it must cope with a consequential computational complexity while controlling for the modifiable areal unit problem (MAUP), spatial autocorrelation and attribute multicollinearity. Nevertheless, we sustain that these effects can reveal significant interactions among diverse spatial phenomena, such as segregation and economic specialization, but most methods treat this apparent disorder as noise. In order to address this issue, we have developed a hierarchical regionalization algorithm that is sensitive to scalar variations of multivariate spatial correlations, recalculating PCA scores at all aggregation steps in order to account for differences in the span of autocorrelation effects for diverse variables. In such a way, we intend to provide a method that minimizes the information loss associated with both MAUP zoning and scale effects, while providing results that allow studying the self organization of spatial patterns avoiding arbitrary zoning decisions. This algorithm produces a hierarchical cartography, which has multiple applications, where two particular cases were studied in Santiago de Chile. With these settings, the scalar evolution of several social distress measures is compared between empirical and 120 random datasets. Remarkably, adjusting several indicators with real and simulated data allows for a clear definition of a stopping rule for spatial hierarchical clustering. Indeed, increasing correlations with scale in random datasets are spurious MAUP effects, so they can be discounted from real data results in order to identify an optimal clustering level, as defined by the maximum of authentic spatial self-organization. This allows to single out the most socially distressed areas in Greater Santiago, thus providing relevant socio-spatial insights from their cartographic and statistical analysis, which agrees to independent diagnostics On the other hand, despite the abundance of works in hedonic mass appraisal, the potential of implementing hierarchical structures to market segmentation has not been fully explored. The purpose of this research is to fill this gap in the literature by studying the impact of incorporating complex architectures to predictive models, such as: econometrics models, artificial neural networks and hybrid models of combined forecasts. Our results confirm that all models exceed their predictive capability when applied in a hierarchical framework In sum, a useful methodology is developed to systematically explore the black box of spatial interdependence and multiscalar self-organizing phenomena, while linking these questions to relevant real world issues.