Fisher判别的基本思想是将k$个总体的所有$维空间的样本点投影到一维空间上，使投影后组与组之间尽可能的分开，然后利用方差分析的方法推出判别函数。为了简单起见，通常利用线性的判别函数。-The basic idea of Fisher discriminant is k $ total $ dimensional space sample point projection to the one-dimensional space, so that the projection between the groups as separate as possible, and then use the method introduced discriminant function analysis of variance. For simplicity, usually using a linear discriminant function.