View article

Globally, 771 million people lack access to electricity, out of which 75% live in Sub-Saharan Africa (IEA, 2020b). Electricity grid expansion can be costly in rural areas, which often have low population densities. Solar home systems (SHS) have provided people worldwide an alternative option to gain electricity access. A SHS consists of a solar panel, battery and accompanying appliances. This research aims to advance the understanding of the SHS customer journey using a case study of SHS customers in Rwanda. This study developed a framework outlining households’ pre- to post-purchase experiences, which included awareness and purchase, both current and future SHS usage and finally customers’ upgrade, switching and retention preferences. A mixed methods approach was utilised to examine these steps, including structured interviews with the SHS providers’ customers (n=100) and staff (n=19), two focus groups with customers (n=24), as well as a time series analysis and descriptive statistics of database customers (n=63,299). A convolutional neural network (CNN) was created to forecast individual SHS users’ future electricity consumption in the next week, month and three months based on their previous hourly usage. Despite the volatility of SHS usage data, the CNN was able to forecast individual users’ future electricity more accurately than the naïve baseline, which assumes a continuation of previous usage. The time series analysis revealed an evening usage peak for non-television users, whilst customers with a television experienced an additional peak around midday. SHS recommendations prior and post-purchase were …