From an initial pool of 1,858 papers in Google Scholar (GS) and 418 papers in Web of Science (WoS), we screened the full texts of 500 papers between 2006 and 2025 and extracted data from 250 studies on both solar irradiance and PV output forecasting. The resulting database contains 5,277 observations across 11 variables, encompassing inputs, methodological, climatic, and contextual factors that differentiate forecast performance. The "r_subset_clean.csv" provides the database.
