Error in my simple code, 3d matrix issue

Here is my code. Basically I am trying to load in 31 matrices then produce one new matrix that is the element by element average of all 31 (So in the new matrix 1:1,1:1 will be the average of the 31 1:1,1:1 values). There is an error somewhere, I think where I make the 2D matrices into 1 3D matrix. The code runs but result is wrong (the matrix that is compiled does not represent the 31 original matrices)

clear all
%Load the Data
S = dir('*.txt');
N = {S.name};
k = 1:numel(S)
A = load(N{k});
Days=zeros(35,43,31);
for k=1:31
Days(:,:,k)=rand(35,43);
end

%Find the Average
Janmean = mean(Days, 3);

%Export the Data
dlmwrite('200301.txt',Janmean)

Thank you in advance for any alterations/help

[Guillaume]

A note on a part of your code that made me pause:

N = somecellarray
k = 1:numel(N); %you use S but N is the same szie
N{k}

is the same as

N = somecellarray
N{:}

The latter shows your intent a lot better

Answer 1

[Jan]

When you omit the evil clear all you can use the debugger to examine your code line by line. The debugger is such useful and required for serious programming, that dleteing all breakpoints is an extremely bad side effect of clear all.

With the debugger you would see, that the load does not do, what you expect. In the loop the values are set to random matrices by the rand command and the loaded data are simply ignored.

    Days = zeros(35,43,31);
  for k = 1:31
    Data = load(N{k});
    Days(:,:,k) = Data.Value;  % Replace "Value" according to you files
  end

[Robert]

Thank you for your reply, however I am a novice and am a bit lost by what to replace "Value" with. Can you give me an example?

[Image Analyst]

clear all does not delete breakpoints if the clear all is in a script, that only happens if you do clear all from the command line. Robert, I do not see the point of loading your mat file into A and then never using A. You probably need to have it inside the for loop like Jan showed you. When you load in the file, all the variables will be "members" or "fields" of a structure, which he called "Data". To get the one particular variable/array that you want, you have to refer to that field. If your variable was called Value, then to extract it from the structure you'd refer to Data.Value. If your 2D matrix variable in the mat file was, say, days2d, then you'd get it by referring to Data.days2d. Does that make sense?

[Robert]

Thanks, but still stumped, if I only put one variable name in the code won't it just use the same one every loop? I have altered the code but still having trouble...

%This code is to calculate the monthly average from daily values in each
%pixel
clear all
%Load the Data
S = dir('*.txt');
N = {S.name};
k = 1:numel(S)
 Days = zeros(35,43,31);
  for k = 1:31
    Data = load(N{k});
    Days(:,:,k) = (Data.???);  
  end

%Find the Average
Janmean = mean(Days, 3);

%Export the Data
dlmwrite('200301.txt',Janmean)

[Image Analyst]

Did you see the FAQ I gave in my answer? It doesn't use the same filename on each iteration.

Answer 2

[Image Analyst]

Attach one of your text files. If they are not .mat format files then you should probably use some other function instead of load(), like csvread(), importdata(), readtable(), textscan() or something like that.

And see the FAQ for how to read a sequence of files: http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

[Robert]

3 Files attached

I'll have a read, thank you

[Image Analyst]

What a mess. Does load() read that in correctly?????

[Robert]

Absolutely, each file loads as a 35x43 matrix. This is the way the data is provided to me

Answer 3

[Guillaume]

Your original code would never work because when you write

load(N{k})

you're passing all the filenames to load at once. load can only open one file at a time. Therefore as per Jan's example your load needs to be inside a loop.

Secondly, I believe that Jan and Image Analyst missed the fact that your load is loading a text file instead of a mat file (or like me assumed it behaved the same for text as for mat, I had to look up the doc). With mat files, load creates a structure whose field names are the variable names stored in the files (hence the replace "Value" according to your file comment). With text files, load just load the data as a matrix.

You were actually nearly there, you'd just need to move the load inside loop and get rid of the rand stuff (what is that about anyway):

%Load the Data
S = dir('*.txt');
numdays = numel(S);
%no need to convert the structure into a cell array
Days=zeros(35,43, numdays); %this is more flexible than hard coding the number of days, particularly if you never check that you actually have 31 files.
for day = 1:numdays  %again avoid hardcoding constants, also use better variable names
    daydata = load(S(day).name);
    %When dealing with user input / data loaded from files, always check that it conforms to your expectations:
    assert(isequal(size(daydata), [35 43]), 'File %s is not a 35x43 matrix', S(day).name);
    Days(:, :, day) = datdata;
end

%Find the Average
Janmean = mean(Days, 3);

%Export the Data
dlmwrite('200301.txt',Janmean)

As per my comments in the code, you're dealing with external input (in this case files) which may not actually be what you expect. For example there may be 32 text files in the directory, or one of the file may not be a 35 x 43 matrix. When you deal with external input, it's always a good idea to check that it conforms to your expectation or not hard code these expectations.