Modeling the Structured Porous Network Using Stacked Ensemble Learning

In this case study, we develop a stacked ensemble learning method to predict the extraction efficiency of the porous network. We generate the porous network data using computational fluid dynamics numerical method. Based on this data we train several machine-learning models. The machine learning models are then used to train a super model on the predictions. Our results show that the stacked ensemble learning is significantly more efficient than every single machine-learning model. The finding of this approach has direct applications in understanding the extraction of oil from porous media.