Classifying Residential Electricity Demand in Mexico using Random Forest and Multinomial Logistic Regression

Conference Paper

This study analyzes the determinants and characteristics that affect the electricity consumption in the Mexican residential sector. This identification is done by applying random forest (RF) and multilinear regression (MLR) methods to find the most influential determinants that explain electricity consumption of the Mexican households. The data used was drawn from the National Survey on the Consumption of Energy Sources in Private Housing Units (ENCEVI) 2018. The results show that differences in socio-demographic and dwelling characteristics are less important than the state in which the households are located. Both methods offered slight results in the validation classification. We believe that this study will be useful for revealing the relevant and irrelevant predictors of electricity demand in other sectors and for particular energy-end uses.

Full Text

Duke Authors

Cited Authors

  • Hernandez, MH; Patiño-Echeverri, D

Published Date

  • December 1, 2019

Published In

  • 2019 Fise Ieee/Cigre Conference Living the Energy Transition, Fise/Cigre 2019

International Standard Book Number 13 (ISBN-13)

  • 9781728142302

Digital Object Identifier (DOI)

  • 10.1109/FISECIGRE48012.2019.8984953

Citation Source

  • Scopus