Hybrid algorithmic methods have become popular to solve complex, cross-disciplinary computational modeling and decision-making problems. This paper suggests a new hybrid algorithm, which combines evolutionary optimization and machine learning-based predictive modelling to enhance the accuracy of solutions, the rate of convergence and the robustness of decisions. The framework was tested on benchmark datasets of engineering design, financial risk assessment and in healthcare decision-making scenarios. The experimental outcomes indicate that the hybrid method is superior to the traditional versions of evolutionary algorithms and individual predictive models, showing an average of 12.5 % improvement in the accuracy of solutions, 18% lower convergence and 9% less computational cost. Also, the sensitivity analysis shows the flexibility of the framework to the levels of complexity of problems, which guarantees the stability of performance in different spheres. Integration of predictive modeling increases the interpretability of the decision and therefore the framework can be used in the real-life scenario where high-stakes decisions are required. On the whole, this work will offer scalable, efficient and interpretable hybrid algorithmic approach which can be used to form the basis of cross-disciplinary computational problem solving.
Paul Ofori-Amanfo (Mon,) studied this question.