Comparative Analysis of Electricity Consumption Patterns Across Four K-Means Clusters of Small and Medium Enterprises

Abstract The paper presents a comparative analysis of the electricity consumption profiles of four consumer clusters identified by k‑means segmentation of hourly smart‑meter data from small and medium enterprises (SMEs) in Latvia. The four clusters were developed using data acquired from 3,016 metering points representing structurally distinct archetypes of temporal consumption behaviour: Cluster 0 (n = 382, evening entertainment), Cluster 1 (n = 1,712, continuously operating infrastructure), Cluster 2 (n = 323, daytime-active retail), and Cluster 3 (n = 599, nocturnal operations). The analysis demonstrates key dimensions of difference across clusters: peak hour of day, day-of-week seasonality, weekend-to-weekday load ratio, diurnal amplitude, and night-time baseload behaviour. These four distinct SME types account for the full range of SME load profile variability observed in the dataset. Implications for demand‑ side management, time‑of‑use tariff design, distribution network planning, and short‑term load forecasting are discussed for each cluster and comparatively across the full typology.