cogent-computing/STAR-Survey-Analysis: Code Release

Since 2003, Brazil is striving to provide energy access to all, in rural areas, in an effort to empower the communities economically. Unpacking fuel stacking behaviour can shed light onto the speed of transition toward the exclusive use of advanced fuel types. This analysis presents findings from surveys carried out with 14 non-electrified communities in a rural area of Rio Negro, Amazonas State, Brazil. We identify the fuel choice determinants in these communities using a multinomial logistic regression model and discuss more generally the validity and robustness of such models in the context of statistical validation and evaluation metrics. Specifically for the Amazonas communities considered in this study, the research showed that fuel choice determinants are the education of household head, the number of females in the household and the number of meals per day in a household. Moreover, considering the actual Brazilian policies related to energy or sustainability is not likely to reach the Sustainable Development Goals proposed by United Nations for 2030. This release contains the analysis and cleaning code applies to the first survey.