Model stacking to improve prediction and variable importance robustness for soft sensor development

Earth and Planetary Sciences

• Model 100%
• Selection 27%
• Prediction 27%
• Proving 18%
• Investigation 9%
• Data Set 9%
• Set 9%
• Need 9%
• Output 9%
• Stacking 9%
• Refinery 9%
• Sensor Development 9%

Computer Science

• Predictive Performance 27%
• Validation 18%
• Interpretability 18%
• Synthetic Datasets 18%
• Soft Sensor 18%
• Ensemble Modeling 9%
• Generalization 9%
• Least Squares Methods 9%
• Random Decision Forest 9%
• Generalization Performance 9%
• Candidate Model 9%
• Individual Model 9%