An association-rule based method for short-term electricity demand forecasting and consumption pattern recognition

Miguel Ángel Zúñiga García

Nombre del Congreso: Mexican International Congress on Artificial Intelligence 2018 (MICAI)

Coautores: Miguel A. Zúñiga-García, G. Santamaria-Bonfil, G. Arroyo-Figueroa, Rafael Batres

Fecha de Presentación: 22/10/2018

Ciudad: Guadalajara

Área del conocimiento: Ingeniería y tecnología - Computación y Ciencias de la Información


Electricity is a commodity that cannot be stored efficiently. Electricity has to be produced at the same time is needed. Electricity demand is highly related to human behaviour. Human behaviour is a rutinary process (Almost every day we eat at the same time, go to work at the same time, go to sleep at the same time, etc.). Association rules is a suitable method to model this kind of behavior. In this work, Association rules method is implemented to analyze 2 year of load data to get specific rules at different moments in a day. The main result is a projection of the most and leas predictable moments in the day to forecast electricity consumption.