This study explores data-driven approaches for modeling industrial processes by employing linear and nonlinear techniques to predict output variables based on available input measurements. Linear regression-based techniques are compared with nonlinear machine learning models to evaluate their predictive capabilities. The analysis considers models trained on high-accuracy & low-frequency laboratory data alongside models leveraging low-accuracy & high-frequency sensor measurements. A hybrid methodology enhances predictive performance by integrating additional process information in the training process. Our findings show that this hybrid approach reduces the RMSE from 0.74 to 0.38 compared to models that rely solely on sensor measurements.