In this report, we present the results of a two-level clustering approach that combines self-organizing maps and K-means to identify distinct household electricity consumption patterns. To address the high dimensionality of the data and better capture both short- and long-term volatility in energy usage, we apply a discrete wavelet transformation to the data expressed in standardized first differences. The resulting transformed dataset serves as input for the clustering methodology. Despite the initial homogeneity of the households in the dataset, four distinct clusters emerge, exhibiting marked differences in energy consumption and intra-day volatility. Moreover, our analysis uncovers a strong positive association between volatility and energy usage: households in clusters characterized by greater intra-day variability consistently consume more electricity. The available socio-economic data further enable us to profile households in the most volatile clusters: they are typically residents of detached houses, rely heavily on electricity for water heating, and
are subscribed to energy plans offering reduced tariffs during nighttime hours and weekends. These findings underscore the importance of targeting consumption volatility in demand-side management strategies. Addressing short-term fluctuations in electricity usage may be a key lever for improving efficiency and reducing overall demand.