(from:)
Mahalanobis distance
In , Mahalanobis distance is a measure introduced by in 1936.It is based on between variables by which different patterns can be identified and analyzed. It gauges similarity of an unknown to a known one. It differs from in that it takes into account the correlations of the and is . In other words, it is a .
Definition
Formally, the Mahalanobis distance of a multivariate vector from a group of values with mean and is defined as:
(注:1.这个是X和总体均值的马氏距离。2.这里的S是可逆的,那么协方差矩阵不可逆的话怎么办?)
Mahalanobis distance (or "generalized squared interpoint distance" for its squared value) can also be defined as a dissimilarity measure between two and of the same with the :
If the covariance matrix is the identity matrix, the Mahalanobis distance reduces to the . If the covariance matrix is , then the resulting distance measure is called the normalized Euclidean distance:
where is the of the ( ) over the sample set.
(源自:百度百科)
马氏优缺点: