Mahalanobis

  马氏距离是一种有效地计算两个样本集之间相似度的算法（数据之间协方差距离），与欧式距离相比，它考虑了各种特征之间的联系。本实验旨在通过给出的样本数据，设计一个最小马氏距离分类器并对测试点进行分类，然后将其与最小欧式距离分类器进行比较，实验得出当协方差矩阵为单位阵时，最小马氏距离分类器将与最小欧式距离分类器等价。本实验还根据实验样本设计贝叶斯分类器进行分类，并与最小马氏距离分类器进行比较，比较两种分类器的分类效果。(Markov distance is an effective method to compute the similarity between the two samples (data covariance distance), compared with the Euclidean distance, which takes into account the link between different characteristics. This experiment aimed at through the given sample data, design a minimum Mahalanobis distance classifier and classify the test points, and then compare it with the minimum Euclidean distance classifier. Experimental results showed that when the covariance matrix is a unit matrix, minimum Mahalanobis distance classifier with the minimum Euclidean distance classifier equivalent.In this experiment, according to the experimental design of the Bias classifier, and compared with the minimum distance classifier, the classification results of the two classifiers are compared.)

Mahalanobis.m,3828,2016-05-24