The global market for smart electricity meters has grown rapidly in recent years and is anticipated to sustain its solid increase in the near future. By analyzing half-hourly meter data from over 4000 Irish households, this study seeks to examine the relationship between households’ attributes and their electricity demand through the following questions: (1) knowing a given set of household attributes, can we accurately classify households according to their demand volume and daily demand pattern and (2) can we infer some of the key households’ characteristics from their meter data. The attributes considered include the size, presence of kids, social class, employment status and the annual income of the households. A range of machine learning methods including tree-based algorithms, support vector machines and neural networks are deployed to answer these questions. The results suggest the potential for reasonably accurate segmentation of consumers according to their demand volume while the classification based on daily demand patterns were shown to be more challenging. For predicting household attributes, higher accuracy values are reported when predicting the household size, social class, and employment status. On the contrary, inferring the household income category and the presence of children in the household were shown to be more difficult.