I got error : Names of fields in structure arrays being concatenated do not match. Concatenation of structure arrays requires that these arrays have the same set of fields.

I have used the following files and the error is :

Error using vertcat Names of fields in structure arrays being concatenated do not match. Concatenation of structure arrays requires that these arrays have the same set of fields.

Error in linearsyntax (line 16) test=[test0;test1];

[Jan]

Please post the complete error message and show us the line, which causes the error. Actually the error message is clear. Before we guess, what you want ti achieve as result, it would be more efficient, if you explain this. Thanks.

[sam CP]

Error using vertcat Names of fields in structure arrays being concatenated do not match. Concatenation of structure arrays requires that these arrays have the same set of fields.

Error in linearsyntax (line 16) test=[test0;test1];

[Image Analyst]

What does this show:

fieldnames(NormalMRI);
fieldnames(AbnormalMRI);

If they don't show the same fieldnames, which is what the error is saying, then exactly what are you trying to do?

Answer 1

[Walter Roberson]

The return value from load() is a structure with one field for each variable stored in the mat file. Even without opening the mat files I can tell that your code instead expects the return value to be the content of a single variable stored in the mat file. You will need to refer to the appropriate field name of the returned structure.

[sam CP]

How can i correct the code ?

[Walter Roberson]

In your code, change the two calls to load() to be calls to load1() and save the following to load1.m

   function r = load1(filename) 
     t = load(filename) ;
     f = fieldnames(t) ;
     r = t.(f{1});

[Walter Roberson]

What is the intention of your lines

training0=NormalMRI(4:1,:);
training1=AbnormalMRI(4:1,:);

test0=NormalMRI(8:5,:);
test1=AbnormalMRI(8:5,:);

In MATLAB, when you have a ":" operator such as 4:1 in which the number before the ":" is larger than the number after the ":", then the result is the empty range.

If your intention is to select rows 1 to 4, you need 1:4 instead of 4:1:

training0=NormalMRI(1:4,:);
training1=AbnormalMRI(1:4,:);

test0=NormalMRI(5:8,:);
test1=AbnormalMRI(5:8,:);

If your intention is to select one row starting from row 4, then you would use

training0=NormalMRI(4:4+1-1,:);
training1=AbnormalMRI(4:4+1-1,:);

test0=NormalMRI(8:8+5-1,:);
test1=AbnormalMRI(8:8+5-1,:);

But then after that you construct group as length 20, in a way that suggests you wanted 10 entries from each in your "training" set. Since your total now of entries in each is 4, my guess is that what you are after is:

NormalMRI= load1('TrainNormalMRI.mat');
AbnormalMRI = load1('TrainAbnormalMRI.mat');

training0=NormalMRI(1:10,:);
training1=AbnormalMRI(1:10,:);

test0=NormalMRI(11:end,:);
test1=AbnormalMRI(11:end,:);

zero=zeros(10,1);
one=ones(10,1);
group=[zero;one];

test=[test0;test1];
training=[training0;training1];

classify(test,training,group)

[IDX,Z]=rankfeatures(training',group);

If you run this then you will get into classify okay, but it will tell you:

The pooled covariance matrix of TRAINING must be positive definite.

You have some work to do in order to correct that. Read https://www.mathworks.com/matlabcentral/newsreader/view_thread/156518

[Walter Roberson]

Note: here, load1() refers to the function I posted above,

 function r = load1(filename) 
     t = load(filename) ;
     f = fieldnames(t) ;
     r = t.(f{1});

[sam CP]

what is "The pooled covariance matrix of TRAINING must be positive definite."

Error using classify (line 228) The pooled covariance matrix of TRAINING must be positive definite.

Error in Test (line 125) classify(test,training,group)

[Walter Roberson]

That is what I linked to, above: https://www.mathworks.com/matlabcentral/newsreader/view_thread/156518 -- you need to read that discussion.

The short short summary is that your do not have enough data samples to get a clean covariance matrix for the number of variables (columns) you are using.

[sam CP]

 function r = load1(filename) 
     t = load(filename) ;
     f = fieldnames(t) ;
     r = t.(f{1});

The code works for my program..Can you explain that how this matlab code works?

[Walter Roberson]

The return value from load() is a structure with one field for each variable stored in the mat file. This code does a load() and queries the field names of the resulting structure and picks the first one. It then does dynamic field name referencing to extract the content of the field.

The net result is to return the content of the "first" variable stored in the .mat file.