1. Introduction
Improved engineering and new business models for electrification have contributed to increasing access to electricity around the world. However, 840 million people still lack access to electricity services [43], many of them residing in places that are difficult to reach and, as a result, expensive to serve [45], [31]. Energy providers, constrained by limited investment budgets, face a perpetual trade-off between expanding electricity access and cost recovery. When consumption levels are low, as can occur in low-income settings, utilities struggle to recover the cost of servicing a grid connection, and the government subsidies[12] for initial capital are poorly utilized. Alternatives to grid extension such as Solar Home Systems (SHS) can support smaller loads without the large wire investments, while in some cases clustered homes (with clusters far from each other) can make mini-grids viable[13]. In practice, identifying those likely to become high consumers is critical to the energy provider, as these are critical to revenue generation and system cost recovery. Given the diversity of electrification technologies, planners rely on energy access planning tools (e.g., the Open Source Spatial Electrification Tool (OnSSET)[27]) that utilize electricity consumption tiers, to match potential customers with technologies that can cost-effectively meet consumption. Consumption predictions can assist matching areas with cost-effective energy technologies, enabling a country to provide electricity access to a larger population given the same investment.