Bronfenbrenner’s Ecological Systems Theory describes child development as a product of nested micro- and macro-systems. At the macro-system level, we consider global phenomena and distal causes that indirectly influence child development. Among them, human-induced climate change has represented a severe threat to human and child development. In this exploratory study, we apply a Support Vector Machine model to rank key environmental aspects of the home and natural environment that predict the Child Development Outcome. We focus on a dataset from Pakistan (N = 21,204 children), extracted from the sixth round of the UNICEF Multiple Indicator Cluster Surveys. The achieved classification performance, measured as Matthew Correlation Coefficient (MCC), was MCC = 0.197 and MCC = 0.179 on the Train and Test partition, respectively. The most important predictor was an environmental measure, the availability of facilities in the home, followed by the presence of physical difficulties. Socioeconomic status (household wealth), but also indicators of the natural context (household owns animals, quality of the drinking water) appeared among the top predictors.
Identifying Key Physical and Natural Environmental Correlates of Child Development: An Exploratory Study Using Machine Learning on Data from Pakistan / Bizzego, A.; Esposito, G.. - 360:(2023), pp. 351-360. [10.1007/978-981-99-3592-5_33]
Identifying Key Physical and Natural Environmental Correlates of Child Development: An Exploratory Study Using Machine Learning on Data from Pakistan
Bizzego A.;Esposito G.
2023-01-01
Abstract
Bronfenbrenner’s Ecological Systems Theory describes child development as a product of nested micro- and macro-systems. At the macro-system level, we consider global phenomena and distal causes that indirectly influence child development. Among them, human-induced climate change has represented a severe threat to human and child development. In this exploratory study, we apply a Support Vector Machine model to rank key environmental aspects of the home and natural environment that predict the Child Development Outcome. We focus on a dataset from Pakistan (N = 21,204 children), extracted from the sixth round of the UNICEF Multiple Indicator Cluster Surveys. The achieved classification performance, measured as Matthew Correlation Coefficient (MCC), was MCC = 0.197 and MCC = 0.179 on the Train and Test partition, respectively. The most important predictor was an environmental measure, the availability of facilities in the home, followed by the presence of physical difficulties. Socioeconomic status (household wealth), but also indicators of the natural context (household owns animals, quality of the drinking water) appeared among the top predictors.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione