ID3-tree.rar_ID3决策树分类_ID3算法_decision tree_id3_信息增益

%计算连续值特征的信息熵值 function [best_AS,midle_data]= C_value_bestgain(C_value,j) Cfn=(1/gama)*2;Cfp=1; [m,n]=size(C_value); C_value_sort=sortrows(C_value,j); %best_gain = 0; best_AS=0 for i= 1:m-1 midle_seris(i,j) = (C_value_sort(i,j)+C_value_sort(i+1,j))/2; C2D_value = C2Dtranlate(C_value,j,midle_seris(i,j)); sy1=find(C2D_value==0); sy2=find(C2D_value==1); yvalue=C_value(:,n); yp=yvalue(sy1); yq=yvalue(sy2); table1=tabulate(yp); [tt1,vv1]=max(table1(:,2)); mp1=table1(vv1); %Nfp=length(find(yvalue~=mp1)); table2=tabulate(yq); [tt2,vv2]=max(table2(:,2)); mp2=table1(vv2); ns=m; %Nfp=length(find(yvalue~=mp2)); if mp1==2 && mp2==2 nn=length(sy1)+length(sy2); C(A)=(nn/ns)*Cfn; elseif mp1==1 && mp2==1 np=length(sy1)+length(sy2); C(A)=(np/ns)*Cfp; elseif mp1==1 && mp2==2 np=length(sy1); nn=length(sy2); C(A)=(np/ns)*Cfp+(nn/ns)*Cfn; elseif mp1==2 && mp2==1 np=length(sy2); nn=length(sy1); C(A)=(np/ns)*Cfp+(nn/ns)*Cfn; end %找到 C2D_value中0和1对应C_value的索引； %根据索引找到相应的Y的值分别存入两个向量； %找到每个向量中值最多的那个标签；并标记选为同一类的元素个数； %建立上述标签矩阵元素并与原始标签元素做比较，找出误分的元素个数，并与相应的代价做乘积；求出C(A) C2D_value1 =[C2D_value,C_value(:,n)]; [m1,n1] = size(C2D_value1); baseentropy = calc_entropy(C_value); feature_value = unique(C2D_value1(:,j)); num_f = length(feature_value); new_entropy = 0; for t= 1:num_f subdata = splitData(C2D_value1 , j, feature_value(t,1)); [m_s,n_s] = size(subdata); prob = m_s./m1; new_entropy = new_entropy + prob * calc_entropy(subdata); end inf_gain = baseentropy - new_entropy; AS=inf_gain*C(A) if AS > best_AS best_AS = AS ; midle_data = midle_seris(i,j); end % if inf_gain > best_gain % best_gain = inf_gain ; % midle_data = midle_seris(i,j); % end end end