Drawing on half-hourly electrical load data from around 1,000 participants in EUDM (now NEAR Program) pilot surveying, these results describe a small set of representative electrical load behavioural classes for Victorian consumers. Each class highlights how behaviour changes over the course of the year and underscores how those behaviours are related to seasons, temperatures and days of the week.

Clustering of houses into groups and groups into representative days is done via unsupervised (automated) clustering that looks for similarities in load shape. No a-priori categorisation is conducted. Labelling of day classes is performed only after the clustering has occurred.

Clustering of homes into groups is as-per the methodology described in the paper by Omid Motlagh (see related assets below).