The rising level of global carbon dioxide (37.5GtCO2e annually) remains a major contributor to climate change, highlighting the urgent need for effective carbon capture and utilization (CCU) strategies. Photocatalysis and electrocatalysis offer promising CO2 reduction routes but are limited by complex reaction mechanisms and low selectivity. In this study, a machine learning (ML) model was developed to predict product yields based on catalyst properties and reaction conditions using literature-derived experimental data. Six ML algorithms including Decision Tree (DT), Random Forest (RF), K-Nearest Neighbors (kNN), Multi-Layer Perceptron Regressor (MLPR), Support Vector Regressor (SVR), and Gradient Boosting Regressor (GBR) were evaluated. Among these, GBR achieved the highest accuracy of 98 %, with predictions generated in under <1 s. Feature importance analysis was used to identify wavelength and light intensity as the most influential parameters. This data-driven approach offers a powerful tool for catalytic systems and accelerates the development of sustainable CO2 utilization technologies.