Operational research encompasses developing and using a wide range of problem-solving techniques and methods applied to improve decision-making and efficiency, such as simulation, mathematical optimization, machine learning, queueing theory, and other stochastic-process models. Operational researchers faced with a new problem must determine which of these techniques are most appropriate given the nature of the system, the goals for improvement, and constraints on time and computing power, or develop a new approach specific to the problem at hand. Nevertheless, operational models only create real value when decision-makers truly start relying on them to optimize their decisions. So, these models should focus on predictive accuracy and scalability, provide insights into data describing past findings, and explain recommendations for decisions. While explainable analytics represents a significant challenge and opportunity for the active research community, the volume of high-quality manuscripts on explainable analytics in journals within operational research is still limited.
The proposed Special Issue “Explainable Machine Learning Operational Applied Research and Applications for Improved Decision-Making” invite high-quality submissions addressing theoretical and algorithmic developments advancing the theory and methodology of explainable Operational Research methods in Mathematical, Computer Systems Science and Engineering as well as real-world innovative implementations in business and society in areas such as biosciences, engineering, marketing and sales, supply chain management, education, production and service operations, medicine, bioinformatics, financial risks, cyber security, and fraud.
Topics for contributions to explainable machine learning/deep learning methods include (but are not limited to):
- Data representation and pre-processing
- Feature engineering and selection methods
- Model-agnostic interpretability methods
- Inherently interpretable algorithms
- Rule-based methods
- Methods for balancing and optimizing predictive performance and interpretability
- Methods supporting model justifiability and actionability
- Privacy-preserving methods
- Methods related to algorithm fairness and bias avoidance
- Interpretable decision-making methods under uncertainty
- Explainable methods for deep learning
- Model visualizations bridging algorithm outcome with domain knowledge
- New model evaluation metrics
- Field tests and real-life experiments that bring analytics closer to the decision-maker
