Efficiency Prediction of E. Conferta in Natural-Sensitized Dye-Sensitized Solar Cells Using k-Nearest Neighbors

Dye-Sensitized Solar Cells (DSSCs) have gained attention as a potential replacement. Recently, natural-sensitized DSSCs have been noted for their low cost and environmental friendliness. However, a significant drawback of natural dyes is their lower efficiency compared to synthetic ones. This has led to extensive research into DSSC characteristics, with Electrochemical Impedance Spectroscopy (EIS) being a prominent focus. Despite its importance, analyzing EIS data can be complex and time-consuming. To address this, machine learning techniques were employed in this study to predict the highest efficiency based on various samples. E. conferta was used as a sensitizer for the DSSCs, and different amounts of Nb-doped TiO2 were tested to create diverse samples. k-Nearest Neighbors (kNN) was utilized to identify the sample with the highest efficiency, revealing that DSSCs with 1.0 wt% Nb-doped TiO2 achieved the highest efficiency, with a prediction accuracy of 90%. To validate the kNN results, manual measurements were conducted, and the findings, presented in Nyquist plots, confirmed that kNN predictions are as reliable as manual measurements for predicting efficiency.