Defining a procedure to identify key sustainability indicators in Spanish urban systems: Development and application

https://doi.org/10.1016/j.scs.2021.102919Get rights and content

Highlights

  • CART and Random Forest are used to identify key sustainable indicators.

  • Three indicators are determined as critical to assess the sustainability of cities.

  • The general predictive accuracy, resulting of the calibration of the model is 87 %.

  • Sustainability of 32 Spanish cities is ranked according with the key indicators.

Abstract

Urban population has exponentially growth in the last decades and as consequence, cities concentrate part of global environmental burdens among other impacts. Therefore, the use of indicators to evaluate the cities in order to achieve their better and more sustainable future was receiving special attention in the last years. Accordingly, the considered indicators would reflect the traditional three pillars of sustainability: social, economic and environmental. In the present study, Classification And Regression Trees (CART) and Random Forest, were applied over a case study in which the sustainability of 31 Spanish cities was evaluated considering 38 indicators. The main goals were to identify the key indicators and to quantify the corresponding thresholds to define a sustainable city. The key indicators identified were: “woman unemployed rate”, “city unemployment rate” and “Municipal Solid Waste collected” and the corresponding thresholds are 14 %, 16 % and 423 kg inhabitant-1 year-1 respectively. In addition, the sustainability of 32 different Spanish cities was evaluated with these three indicators and thresholds to validate the achievements. According with the results, urban sustainability could be evaluated considering only three indicators with a high degree of accuracy, providing information to policy makers without the requirement of compiling a large amount of data.

Introduction

The accelerated urban population growth in the last century has resulted in cities that constitute an important economic source generating more than 80 % of the global gross domestic product (GDP). At the same time, however, cities play a predominant role in the consumption of natural resources, as well as being a major actor in the impacts on the three environmental compartments: air, water and soil (Kennedy, Cuddihy, & Engel-yan, 2007; World Bank, 2019). Based on population projection studies for the next two decades, it is foreseeable that 60 % of the total world population will live in cities (John, Luederitz, Lang, & von Wehrden, 2019), with consumption profiles higher than the earth's capacity to provide natural resources (Goldstein, Birkved, Quitzau, & Hauschild, 2013). The displacement of the population from low-productivity rural and agricultural areas to large cities such due to economic expansion may, on the other hand, increase inequality in the distribution of income, so that unbalanced growth could affect citizens in terms of unemployment rates, social inequality, poverty and/or cost of living (Feleki, Vlachokostas, & Moussiopoulos, 2018). The concern for sustainability assessment in cities and/or urban systems is reflected in the Sustainable Development Goals (SDGs) of the United Nations, such as SDG-11, whose main target is to make cities and human settlements inclusive, safe, resilient and sustainable (United Nations, 2015). Kennedy et al. (2007) defined a sustainable urban system as one that consumes resources and generates waste without exceeding the capacity of nature to regenerate more resources and assimilate that waste. Nevertheless, there is no agreement between researchers and policymakers on the definition of “sustainability” when applied to urban systems.

The use of sustainable development indicators has become popular in the last decade among public administrations and policy makers as a way of identifying the degree to which the proposed objectives have been met and, from there, to improve strategies and action plans (Tanguay, Rajaonson, & Lanoie, 2010). Nevertheless, the number of available indicators is quite high (Feleki et al., 2018), and it is necessary to reduce this number to a more manageable set of indicators (González-García et al., 2019). On the other hand, depending on the study, the set of indicators selected varies according to the evaluator’s point of view (Tanguay et al., 2010).

For this reason, the selection of the indicators set is a critical step in the sustainability assessment. There are different international standard organizations, such as International Organization for Standardization (ISO), European Telecommunication Standards Institute (ETSI) and Telecommunication Standardization Sector (ITU), which provide and recommend different indicators to assess the sustainability of urban systems (International Standardization Organization, 2014). These recommended indicators could be considered as a starting point, which, according to the needs of each city, includes a more manageable set that provides the information required for the city under study (Huovila, Bosch, & Airaksinen, 2019). However, the most common application is the comparison between cities in terms of sustainability (Tanguay et al., 2010). Accordingly, if each study considers different indicators for evaluation, such comparison will not be possible. In this sense, it is necessary to develop methodologies focused on the selection of potential indicators to assess the sustainability of cities in a more holistic and systematic way (Feleki et al., 2018).

In addition, the comparative analysis between cities provides information on the shortcomings of cities with lower indicator values and, on the other hand, makes it possible to identify the cities with better results, which are usually linked to sustainability policies (Rozhenkova et al., 2019). For example, if one city has an unemployment rate of around 20 % but another city in the same country has around 10 %, this implies that the policy makers of the first city must orient their employment policies and strategies according to those of the second city. The main difference between comparative and individual analysis is that in the latter it is difficult to know whether 20 % of the unemployment rate, following the example, is a bad or a good value. In this way, reference values or thresholds could be calculated and established in the comparative analysis within a sample of the cities studied. Moreover, the threshold may help to define the criteria under which a city is established whether or not it is considered sustainable (González-García et al., 2019). However, the comparative analysis of cities requires an exhaustive compilation of data. It is therefore necessary to collect information at city level for all selected indicators.

In this context, some authors developed and applied different methodologies to evaluate the sustainability of cities using indicators. Some studies proposed a set of indicators taking into account the specific characteristics of the case study, such as the study of peripheral informal settlements in Latin American cities (Montoya, Cartes, & Zumelzu, 2020), or the study of the strengths and weaknesses of a specific area (Hély & Antoni, 2019) or even the study of some characteristic sector, such as tourism (Biagi, Ladu, Meleddu, & Royuela, 2020). However, the number of indicators could be low (no more than 10) and very specific to the case study. In this regard, there are other studies in which the selection of the set of indicators was first introduced based on databases, regardless of the characteristics of the urban system. It is necessary to incorporate screening methods to identify indicators that are representative of each case study in order to establish specific criteria for the selection of key indicators (Bienvenido-Huertas, Farinha, Oliveira, Silva, & Lança, 2020; Chen & Zhang, 2021; González-García et al., 2019; Karji, Woldesenbet, Khanzadi, & Tafazzoli, 2019). Consequently, the set of indicators identified in these studies can be easily adapted and used in the analysis of other case studies (Rama, González-García, Andrade, Moreira, & Feijoo, 2020). However, despite reducing the number of indicators, these studies required a large amount of data to be compiled, which requires time and resources.

The main objectives of this study are to develop a methodology to predict the sustainability of a Spanish city with only three indicators and to apply the methodology developed on a case study. For this purpose, two analysis models (Classification and Regression Trees and Random Forest) were applied on an initial case study of Rama et al. (2020) composed of a sample with the 31 most representative Spanish cities and 38 indicators. Once the methodology was developed from the case study as a starting point, it was applied to a new case study composed of the 32 Spanish cities with more than 100,000 inhabitants that were not considered in the initial stage. Consequently, the novelty of this study is the use, for the first time, of these analysis models to identify key indicators for assessing sustainability in urban systems without losing accuracy.

Consequently, this study has been structured in five stages: 1) description of the analysis models; 2) description of the case study as a baseline reference; 3) identification of key sustainability indicators; 4) calibration on the aforementioned case study and, finally, 5) application to a new sample, with the objective of determining its sustainability ranking.

Section snippets

Materials and methods

The analysis methodology consists of two analysis models applied to a sample of 31 cities in which 38 indicators were previously measured, divided into 14 social, 13 economic and 11 environmental indicators. Bearing in mind which cities were sustainable and which were not, according to the values reached in the indicators, the two analysis models predicted which indicators had the greatest predictive capacity to determine the sustainability profile of each city. Moreover, they made it possible

Results and discussion

This section has been divided into three sub-sections taking into account the main objectives of this study. First, the key indicators and corresponding thresholds resulting from the application of the analysis models were identified. In this sense, the set of 38 indicators was reduced to a set of keys. Secondly, the sustainability scores of these cities were again recalculated using only the key indicators and the corresponding thresholds to calibrate the method. Finally, a new case study

Conclusions

Concern about the sustainability of urban systems and cities has been growing in recent decades. Within the definition of a sustainable city, in addition to environmental factors, social and economic ones must also be considered. For this reason, the use of the analysis of the sustainability of urban systems with indicators grew in recent years because it allows evaluating sustainability from the point of view of these three factors. However, there is a lack of consensus when it comes to

Declaration of Competing Interest

All authors have participated in (a) conception and design, or analysis and interpretation of the data; (b) drafting the article or revising it critically for important intellectual content; and (c) approval of the final version.

Acknowledgements

This research was supported by a project granted by the Spanish Government and FEDER/ Ministry of Science and InnovationSpanish National Research Agency (CTQ2016-75136-P) and by a project granted by Xunta de Galicia (project ref. ED431F 2016/001). Dr. S.G.-G. would like to express her gratitude to the Spanish Ministry of Economy and Competitiveness for financial support (Grant reference RYC-2014-14984). The authors belong to the Galician Competitive Research Group GRC ED431C 2017/29 and to

References (52)

  • B. John et al.

    Toward sustainable urban metabolisms. From system understanding to system transformation

    Ecological Economics

    (2019)
  • E.A. Kanin et al.

    A predictive model for steady-state multiphase pipe flow: Machine learning on lab data

    Journal of Petroleum Science & Engineering

    (2019)
  • A. Karji et al.

    Assessment of social sustainability indicators in mass housing construction: A case study of Mehr housing project

    Sustainable Cities and Society

    (2019)
  • I. Kiaušienė

    Comparative assessment of women unemployment and poverty in European Union

    Intellectual Economics

    (2015)
  • M.A. Krebs et al.

    Predicting understory vegetation structure in selected western forests of the United States using FIA inventory data

    Forest Ecology and Management

    (2019)
  • H. Li et al.

    Predicting business failure using classification and regression tree: An empirical comparison with popular classical statistical methods and top classification mining methods

    Expert Systems with Applications

    (2010)
  • M. Rama et al.

    Assessing the sustainability dimension at local scale: Case study of Spanish cities

    Ecological Indicators

    (2020)
  • A. Rico et al.

    Carbon footprint of tourism in Barcelona

    Tourism Management

    (2019)
  • V. Rozhenkova et al.

    The role of comparative city policy data in assessing progress toward the urban SDG targets

    Cities

    (2019)
  • G.A. Tanguay et al.

    Measuring the sustainability of cities: An analysis of the use of local indicators

    Ecological Indicators

    (2010)
  • G.N. Zaimes et al.

    Assessing the impact of dams on riparian and deltaic vegetation using remotely-sensed vegetation indices and Random Forests modelling

    Ecological Indicators

    (2019)
  • J.B. Ahmed et al.

    Spatial assessment of termites interaction with groundwater potential conditioning parameters in Keffi, Nigeria

    Journal of Hydrology

    (2019)
  • Ais Group

    Spanish poverty town to town

    (2016)
  • B. Biagi et al.

    Tourism and the city: The impact on residents’ quality of life

    International Journal of Tourism Research

    (2020)
  • D. Bienvenido-Huertas et al.

    Comparison of artificial intelligence algorithms to estimate sustainability indicators

    Sustainable Cities and Society

    (2020)
  • L. Breiman

    Random forests

    Machine Learning

    (2001)
  • Cited by (0)

    View full text