Wasserstein Distributionally Robust Optimization in Multivariate Ridge Regression

Distributionally robust optimization is an effective method to deal with uncertainty. In this paper, we apply distributionally robust optimization methods to multivariate ridge regression model with data uncertainty. We construct the ambiguity set of the distributionally robust optimization problem by Wasserstein metric. We relax the primal formulation of the distributionally robust multivariate ridge regression model by introducing a regularization term. In this approach, the relax formulation is a solvable convex optimization problem. The feasibility and effectiveness of our approach is verified by numerical experiments.