Ultrapure water (UPW) is essential in industries like semiconductor manufacturing, where demand is rising. Raw water used for UPW production often contains low molecular weight organic compounds, which are difficult to control. In the VUV/UV process, lamps emitting 185 nm and 254 nm wavelengths degrade dissolved organic carbon (DOC). However, the 185 nm light also generates hydroxyl radicals by decomposing water, leading to hydrogen peroxide (H₂O₂) formation, which can adversely affect semiconductor product quality. Therefore, understanding and controlling H₂O₂ formation in the VUV/UV process is critical. In this study, nitromethane, acetonitrile, methanol, and isopropyl alcohol were selected as representative low molecular weight organic compounds commonly found in the final effluent of UPW production processes. We investigated how the concentrations of these compounds, pH variations, and hydraulic residence time (HRT) influence H₂O₂ formation. The results show that lower pH and higher organic compound concentrations increase H₂O₂ formation. A longer HRT also promotes more radical generation, leading to higher H₂O₂ formation. A machine learning model was developed to estimate hydroxyl radical concentrations during the VUV/UV process and predict H₂O₂ formation. The model confirmed that higher HRT, dissolved oxygen (DO), and compound concentrations, along with lower pH, result in increased H₂O₂ levels, aligning with experimental data. This model is expected to enhance UPW production by enabling accurate predictions of H₂O₂ concentrations during the VUV/UV process.