A methodology to classify residential customers by their pattern of consumption

In Mexico, in 1996, daylight-saving time (DST) was implemented during the Summer (April to October). In order to evaluate the savings due to DST, 600 residential customers were monitored. Since the saving is not a function of the consumption range, a mathematical model was developed, which allowed classification of the customers, by means of their consumption profile called patterns of consumption (PC). In this paper, the authors present the procedure that was used to obtain the PCs. It is related with the multivaried statistics method named cluster analysis. The PCs for each city were obtained in four stages: (1) determination of the differences among the consumptions of all customers; (2) calculation of the distances between the profiles of each customer; (3) determination of the number of groups (typical profiles of a collection of customers); and (4) determination of the percent representation by group, of the total sample. The effect on various types of customers and in various climatic conditions is also shown.