Modeling Electricity Consumption through Socioeconomic Indicators: A Multivariable Regression Approach

This study examines the relationship between electricity consumption and sociodemographic factors across 258 municipalities in Honduras using data from national statistical sources. Multiple regression models were simulated in RStudio to identify the main predictors of municipal electricity demand. The initial linear model achieved high explanatory power (R2=0.85), but diagnostic tests revealed heteroskedasticity and non-normal residuals, limiting the model's reliability. To correct these issues, a log-linear model using the natural logarithm of electricity consumption was developed. This specification eliminated heteroskedasticity (Breusch-Pagan p = 0.282) and produced a more stable fit (R2 = 0.618; R2 test = 0.81). Results indicate that urbanization, income per capita, and human development positively influence electricity consumption, while poverty level has a negative effect. This last model provides a tool to increase the decentralized power system planning since municipalities can better forecast their electricity demand and allocate resources (investments in distribution or generation) when and where they need them.