Instance-Based Learning-Driven Density of States Analysis in Functionalized Fullerene Derivatives for Optimizing Organic Photovoltaics

We employ instance-based machine learning to evaluate functionalized fullerene derivatives as electron-acceptor materials for organic photovoltaics. Using density of states (DOS) data from seven fullerene-based models, we identify key electronic properties that influence photovoltaic efficiency. Our results demonstrate that k-nearest neighbor methods achieve superior predictive accuracy, underscoring the potential of instance-based learning in accelerating material discovery and optimization for next-generation photovoltaics.