A Bias-Variance-Complexity Trade-Off Framework for Complex System Modeling

This study proposes a new complex system modeling approach by extending a bias-variance trade-off into a bias-variance-complexity trade-off framework. In the framework, the computational complexity is introduced for system modeling. For testing purposes, complex financial system data are used for modeling. Empirical results obtained reveal that this novel approach performs well in complex system modeling and can improve the performance of complex systems by way of model ensemble within the framework.