cal_difference1.m

两种改进的遗传算法（自适应交叉概率的遗传算法

%计算两类间的方差和类内方差
%假设某一类样本数为20，与该类有关的特征为10，则该类的矩阵为20×10，20为行数，10为列数
%Name:cal_difference1.m

function [difference]=cal_difference(M1,M2)
double var_sum 
[px_m1,py_m1]=size(M1);
[px_m2,py_m2]=size(M2);
vector_difference=zeros(py_m1,1);
for j=1:py_m1
    mean_m1=mean(M1(:,j));
    mean_m2=mean(M2(:,j));
    mean_m1_m2=(mean_m1-mean_m2)^2; %求解两矩阵间某一列向量间的均值之差,同个特征两类差异越大越好
    vector_var=var(M1(:,j))+var(M2(:,j));%求解两矩阵间某一列向量间的方差之和，同个特征同类差异越小越好
    vector_difference(j,1)=mean_m1_m2/vector_var;
end
difference=vector_difference;%25 hang 1 lie