Gradient Boosting Algorithm to Identify Markets for Residential Solar: New York State Case Study

摘要

It is of crucial importance for commercial entities (e.g., power utilities, developers, and even governments) to identify profitable markets for residential solar PV installations. This study estimates the potential number of customers and the expected market size for residential solar panels in the state of New York. The study uses a gradient boosting model with sample weights to estimate the probability of a household installing solar in a given census tract. The features of the model are a combination of household demographics, housing stock characteristics (e.g., number of rooms), potential solar electricity generation, and electricity market prices. Results show that the potential number of customers for new solar panel installations in New York State is 78,700 with an expected market size of $3,590 millions.