Titolo della tesi: Statistical Learning Methods to synthetise Sustainable Development Goals and forecast progress towards targets
This dissertation reflects on the impact of the COVID-19 on the SDGs, on both nowcasted (2020) and forecasted (2030) indicators in which the FAO is the custodian agency with a Double Exponential Smoothing approach. The analysis shows that the pandemic will have dramatic consequences, eroding many of the gains recorded over the last decade in terms of food security, sustainable agriculture, and biodiversity (SDGs 2, 6, 14, and 15). The analysis demonstrates that the indicators are interlinked and have a significant endogenous factor. The indicators have a hierarchical design form and complex system. This characteristic allows for summarising the complexity, thus reducing the number of indicators. The analysis employed a combination of traditional statistical methods and applied econometrics including higher-order factor analysis, Structural Equation Modelling, Ordinary Least Squares, Least Absolute Shrinkage and Selection Operator, and Three-Stage Least Squares. These findings suggest that targeting policies to the most influential indicators can be most effective. This not only prioritises resources but also influences the interlinked indicators towards a more conducive investment for achieving the SDGs.