<p>Perovskite solar cells have been under extensive investigation for the past few decades as they promise higher efficiency and lower cost of production compared to current silicon-based solar cells. The implementation of any perovskite either in single or multijunction solar cells is mainly dependent upon its energy bandgap. The existing methods to determine and predict the bandgap of a perovskite are time-consuming, expensive and resource intensive. In this work, we discuss leveraging machine learning algorithms and techniques to identify the key compositions influencing the bandgap of lead (Pb) and tin (Sn)-based perovskite solar cells. For Pb-based perovskite solar cells, Cs_aFA_bMA_(1-a-b)Pb(Cl_xBr_yI_(1-x–y))₃ configuration has been considered to predict the impact of composition on the bandgap by applying various machine learning models. Similarly, for Sn-based perovskite solar cells, we have investigated Cs_aFA_bMA_(1-a-b)Sn(Cl_xBr_yI_(1-x–y))₃ configuration to precisely make the bandgap prediction. The machine learning models are applied for both the configurations by considering 80:20 ratio for trained and tested datasets. For Pb-based perovskite solar cells, the neural network model predicted the bandgap with highest accuracy, whereas the ExtraTreeRegressor model performed best for predicting the bandgap of Sn-based perovskites. These findings demonstrate the potential of machine learning to accelerate the development of high-efficiency, cost-effective perovskite materials, offering a transformative approach for the photovoltaic industry in its shift toward next-generation solar technologies.</p>