Index exceeds matrix dimensions & Dimensions of matrices being concatenated are not consistent.

hello everybody,

i am sharing my code and its giving some errors, i hope any of you guys will try to help me. here are 2 .m files. one is RM.m, in this file i am calling a function Train_Test.

//RM.m

// ind_pin.mat fill consist of 2 sub matrix, "fea" and "gnd", "fea" is 200x21025 and "gnd" is 1x21025.

clear;
clc
load ind_pin.mat 
for kk=1:3
    x = sprintf('iteration = %d',kk);
    disp(x);
    for numTrain=2:4
         for M=5:2:7
             L=20;
             [feaTrain,gndTrain,feaTest,gndTest]=Train_Test(fea,gnd,numTrain);

               [eigvector_PCA, eigvalue_PCA] = PCA(feaTrain,0);
               [eigvector_Select] = RandomSelect(eigvector_PCA, 140);
               pca_feaTrain = feaTrain*eigvector_Select;
               pca_feaTest = feaTest*eigvector_Select;

               acc=Accury(pca_feaTrain,gndTrain,pca_feaTest,gndTest,M,L);
               A = sprintf('Accuracy = %6.3f',acc);
               disp([A '%']);
           end
      end

end

/Train_Test.m 

//function
function [feaTrain,gndTrain,feaTest,gndTest]=Train_Test(fea,gnd,numTrain)
X = sort(gnd);
temp = diff([X;max(X)+1]);
count = diff(find([1;temp]));
y = [X(find(temp)) count];
clear X temp count;

[nClass, tmp] = size(y);
 m = 1;
    for i = 1 : nClass
        cSampleIdx = find(gnd == y(i,1));
        R = randperm(y(i,2));
        for j = 1 : numTrain
            feaTrain((i-1)*numTrain+j,:) = fea(cSampleIdx(R(j)),:);
            gndTrain((i-1)*numTrain+j) = gnd(cSampleIdx(R(j)));
        end
        for j = numTrain+1 : y(i,2)
            feaTest(m,:) = fea(cSampleIdx(R(j)),:);
            gndTest(m) = gnd(cSampleIdx(R(j)));
            m = m+1;
        end
    end
    gndTrain = gndTrain';
    gndTest = gndTest';
    clear R i j m cSampleIdx;

when i run it gives me following errors.

"Error using vertcat
Dimensions of matrices being concatenated are not consistent.
Error in Train_Test (line 4)
temp = diff([X;max(X)+1]);
Error in RMob (line 25)
[feaTrain,gndTrain,feaTest,gndTest]=Train_Test(fea,gnd,numTrain);

Index exceeds matrix dimensions.
Error in Train_Test (line 15)
feaTrain((i-1)*numTrain+j,:) = fea(cSampleIdx(R(j)),:);
Error in RMob (line 25)
[feaTrain,gndTrain,feaTest,gndTest]=Train_Test(fea,gnd,numTrain);

Answer 1

[Walter Roberson]

You have

X = sort(gnd);
temp = diff([X;max(X)+1]);

Your gnd is a row vector, so sort() of that will be a row vector, so X is a row vector. max(X) applied to a row vector is a scalar, and adding 1 to a scalar gives a scalar, so max(X)+1 is a scalar. X is still a row vector. So you have [row_vector; scalar] which tries to place the scalar "underneath" (on the next row) from the row vector. But the scalar does not have the same number of columns as the row vector does, so it fails.

You need to decide whether your [X;max(X)+1] needs to become a row vector or a column vector and make the appropriate change to the code.