RESUMO
In a companion article, previously published in Royal Society Open Science, the authors used graph theory to evaluate artificial neural network models for potential social and building variables interactions contributing to building wind damage. The results promisingly highlighted the importance of social variables in modelling damage as opposed to the traditional approach of solely considering the physical characteristics of a building. Within this update article, the same methods are used to evaluate two different artificial neural networks for modelling building repair and/or rebuild (recovery) time. By contrast to the damage models, the recovery models (RMs) consider (A) primarily social variables and then (B) introduce structural variables. These two models are then evaluated using centrality and shortest path concepts of graph theory as well as validated against data from the 2011 Joplin tornado. The results of this analysis do not show the same distinctions as were found in the analysis of the damage models from the companion article. The overarching lack of discernible and consistent differences in the RMs suggests that social variables that drive damage are not necessarily contributors to recovery. The differences also serve to reinforce that machine learning methods are best used when the contributing variables are already well understood.
RESUMO
The use of machine learning has grown in popularity in various disciplines. Despite the popularity, the apparent 'black box' nature of such tools continues to be an area of concern. In this article, we attempt to unravel the complexity of this black box by exploring the use of artificial neural networks (ANNs), coupled with graph theory, to model and interpret the spatial distribution of building damage from extreme wind events at a community level. Structural wind damage is a topic that is mostly well understood for how wind pressure translates to extreme loading on a structure, how debris can affect that loading and how specific social characteristics contribute to the overall population vulnerability. While these themes are widely accepted, they have proven difficult to model in a cohesive manner, which has led primarily to physical damage models considering wind loading only as it relates to structural capacity. We take advantage of this modelling difficulty to reflect on two different ANN models for predicting the spatial distribution of structural damage due to wind loading. Through graph theory analysis, we study the internal patterns of the apparent black box of artificial intelligence of the models and show that social parameters are key to predict structural damage.
