In order to evaluate electricity consumption as a socioeconomic indicator, this research evaluates the connection between electricity consumption and the household income of those enrolled in Brazilian social programs (usually poor families). The research aims to contribute to a more accurate socioeconomic classification of the population in need of assistance programs and other public policies to overcome social inequalities, as well as to support strategic decisions of the Public Administration. Regarding the methodology, the research is based, especially, on the microdata of the Single Registry for Federal Government Social Programs (called CadÚnico) of the State of São Paulo, located in the Brazilian Southeast region, coupled with the residential consumers' database of the Social Tariff of Electrical Energy (TSEE). This research applied techniques from multiple linear regressions (OLS) towards Big Data Analytics and Data Science techniques through methods of statistical inference: spatial autocorrelation (Moran´s I), Spatial Auto-Regressive (SAR) models and Geographically Weighted Regression (GWR).