Comparing Hyper-optimized Machine Learning Models for Predicting Efficiency Degradation in Organic Solar Cells