This work deals with prediction of power conversion efficiency and short circuit current of an ITO/TiO2/CsSnI3/Cu2O/Au solar cell using four machine learning (ML) models namely decision tree (DT), random forest (RF), gradient boosting (GB) and XGBoost. The solar cell device is designed in SCAPS and achieved an optimized short-circuit current density of 34.75 mA/cm2, open-circuit voltage of 1.09 V, and fill factor of 82.97%, leading to an impressive PCE of 30.85% at perovskite thickness of 1000 nm, doping density of 1 × 1019 cm−3, interface defect densities of 1 × 1010 cm−3 and the bulk defect density of 1 × 1010 cm−3. An in-house dataset consisting of 20,000 data for the device is developed using SCAPS and used to train the ML models. Amongst the MLs the XGBoost showcases the best with an accuracy of 99.99%. Out of the five features of the device the contribution of interface defect, bulk defect and doping of perovskite to train the ML models is found to be high. Contrarily the thickness of the perovskite strongly opposes the accuracy of the models. This work showcases a simple method to reduce the time-consuming proof of concept step during the development of new perovskite solar cell.