Targeting SDG7: Identifying heterogeneous energy dilemmas for socially disadvantaged groups in India using machine learning

To achieve Sustainable Development Goal (SDG) 7, prioritizing the socially disadvantaged segments of the population is imperative, given their inherent susceptibility to heightened risks of energy exclusion. However, a comprehensive understanding of the diverse energy challenges faced by households with socio-economic disparities remains elusive. This article thus addresses this gap by examining three widely acknowledged categories of marginalized households in India: racial inferiority, income poverty, and gender inequality. It notably pioneers the quantification of an umbrella pattern of energy deprivation within the SDG7 framework, encompassing energy unaffordability, energy unreliability, energy inaccessibility, and energy inequality. To do so, leveraging the latest household survey dataset and employing least squares estimates, we preliminarily capture that these three disadvantaged groups encounter significant energy barriers in the pursuit of SDG7 achievement. Given respectively selected models based on Least Absolute Shrinkage and Selection Operator (LASSO) approach, the gradient boosting model (GBM), another state-of-the-art machine learning technique, is subsequently adopted to verify feature significance and rank its importance in determining diverse energy deprivation faced by each group. The results reveal that the disadvantaged caste groups and those experiencing greater gender inequality encounter the greatest impediments to their right to reliable energy access. In comparison, energy unaffordability poses a paramount challenge for low-income households. These findings enable policymakers to design straightforward interventions that address a spectrum of socio-economic disparities, thereby fostering an just energy transition grounded in data-driven evidence.