1: \begin{abstract}

2: Rapidly growing product lines and services require a finer-granularity forecast that considers geographic \regions. 

3: However the open question remains, how to assess the quality of a spatio-temporal forecast?

4: In this manuscript we introduce a metric to evaluate spatio-temporal forecasts. 

5: This metric is based on an Optimal Transport (OT) problem. 

6: The metric we propose is a constrained OT objective function using the \emph{Gini impurity function} as a regularizer. 

7: We demonstrate through computer experiments both the qualitative and the quantitative characteristics of the Gini regularized OT problem. 

8: Moreover, we show that the Gini regularized OT problem converges to the classical OT problem, when the Gini regularized problem is considered as a function of $\lambda$, the regularization parameter. 

9: The convergence to the classical OT solution is faster than the state-of-the-art Entropic-regularized OT\citep{cuturi2013sinkhorn} and results in a numerically more stable algorithm.

10: \end{abstract}

11: