Machine Learning Reduces Soft Costs For Residential Solar Photovoltaics

This study suggests a shift from significance-based models to prediction-oriented methods, and demonstrates that machine learning algorithms can improve the prediction performance for PV adoption. It can also lower PV companies’ customer acquisition costs, make the technology more affordable, and further increase the market size of the industry. Our research findings can improve the decision-making of utility planners and policy-makers by providing a better prediction of the location and market size of future PV adoption.