Applying an Alternative Machine-Learning Qualification Metric: Impacts for Inclusion, Accuracy and Revenue

Financing for solar products often requires traditional credit scores to determine risk metrics, limiting renewable energy access for millions of low-credit or credit invisible households, disproportionately low-to-moderate income (LMI) and minority communities.

Improving upon traditionally relied upon credit scores, we apply a machine-learning qualification metric to historical community solar customer application data to determine the extent to any gains in accuracy, inclusion and revenue.

By using a proprietary qualification tool that more effectively quantifies risk, this research demonstrates the ability to increase access to solar energy financing, particularly for LMI populations, while decreasing aggregate risk of default.

Solstice operates EnergyScore, a machine learning algorithm that provides an alternative assessment for utility bill default risk. As increasing attention is placed upon equitable renewable access, this research will provide timely lessons of applying an advanced algorithm alternative in the field. This topic is relevant to machine learning applications in renewable financing, LMI inclusion and equity, and implications on revenue generation. This research provides lessons learned for applications of alternative credit metrics, useful for organizations with mission-oriented goals of increasing access, particularly for LMI populations.