How we identified similar groups of households based on energy consumption
To group households effectively, it's essential to decide which characteristics will be used for the grouping. With this in mind, grouping households according to the heating system they own will likely result in a different set of groups compared to grouping households based on the number of people in the household.
An energy-use profile is a group of households that have similar patterns of energy usage. Before identifying energy-use profiles, our multidisciplinary team outlined factors (behaviours, lifestyles and property characteristics) that might contribute to differences in energy consumption patterns. These were then mapped to features we could extract from half-hourly gas and electricity consumption data – what we call smart meter features.
Factors and behaviours that influence energy consumption and how to measure them
To take a look at the complete list of factors, behaviours and respective features, please read the technical appendix.
The range of smart meter features we defined and depicted above (such as average overnight electricity consumption) helps create a fuller picture of household energy consumption than is typically available.
We applied a clustering algorithm – a data science technique used to group data points into clusters based on how close they are to each other – to our dataset of smart meter features to find groups of similar households. For each of these groups, patterns of energy consumption across the day, year and seasons can be visualised. The resulting energy-use profiles help us to see beyond simple accounts of average annual consumption, while still representing coherent groupings of GB households. For example, if you want to help households flex their energy consumption, then you might need to know when they consume energy, but not necessarily how much they consume. You can read more about applications in the section at the end.
