Proposing a composite index and maturity model for urban sustainability in the Brazilian context: A machine learning and data envelopment analysis approach

Abstract

AbstractDespite efforts towards the accomplishment of the UN 2030 Agenda, the challenges of burgeoning populations, income inequality, difficulty in accessing basic services, among others, remain in several cities around the world. New approaches to measure and assess the sustainability of cities can support the development of actions to improve the different dimensions of sustainability. The research aims to propose an urban sustainability index and a maturity model to evaluate the sustainability of cities and monitor it over time. To achieve this objective, a maturity model was developed based on three different techniques: Data Envelopment Analysis, Artificial Neural Networks, and Analysis of Variance. The proposed index and the maturity model were applied to evaluate a sample of 504 Brazilian cities. The main results observed are: (a) the presence of five distinct levels of city performance (maturity), grouped via machine learning and validated via inferential statistics; (b) no city was considered fully sustainable and only 4.76% of the cities studies are at the highest level of urban sustainability maturity; (c) from a joint application of the three quantitative techniques and specific targets for each indicator could be identified, and the performance of cities classified over time. Based on the results, it is hoped that policy makers will have more objective and standardized tools to collect useful information and be able to reinforce critical strategies or chart new policies towards sustainable urban development. It is also hoped that the joint application of the techniques can shed light on new urban sustainability assessment models.