Mahalanobis Distance

Measures distance between an object xx and the Distribution of the dataset.

MD⁡(x)=(x−x‾)TC−1(x−x‾)\operatorname{MD}(\mathbf{x})=\sqrt{(\mathbf{x}-\overline{\mathbf{x}})^T \mathbf{C}^{-1}(\mathbf{x}-\overline{\mathbf{x}})}

with x‾\overline{x} Arithmetic Mean and CC the Covariance matrix.

Use Grubbs Test on this measure to detect outliers.