The adoption of residential-scale low carbon technologies, such as photovoltaic panels or electric vehicles, is expected to significantly increase in the near future. Therefore, it is important for distribution network operators (DNOs) to understand the impacts that these technologies may have, particularly, on low voltage (LV) networks. The challenge, however, is that these LV networks are large in number and diverse in characteristics. In this work, four clustering algorithms (hierarchical clustering, $k$ - ${\rm medoids}$ ++, improved $k$ - ${\rm means}$ ++, and Gaussian Mixture Model—GMM), are applied to a set of 232 residential LV feeders from the North West of England to obtain representative feeders. Moreover, time-series monitoring data, presence of residential-scale generation, and detailed customer classification are considered in the analysis. Multiple validity indices are used to identify the most suitable algorithm. The improved $k$ - ${\rm means}$ ++ and GMM showed the best performances resulting in eleven representative feeders with prominent characteristics such as number and type of customers, total cable length, neutral current, and presence of generation. Crucially, the results from studies performed on these feeders can then be extrapolated to those they represent, simplifying the analyses to be carried out by DNOs. This is demonstrated with a hosting capacity assessment of photovoltaic panels in LV feeders.