how to get average of multiple rows and subtract from one group of average to another

Hello,

I have a big file with multiple rows and columns. I need to get the average of every 30 rows. for example from row 1 to 30, 31 to 60, 61 to 90 and so on until the end of the rows. Can anyone help me with this? any kinds of help will be really appriciated.

[Shlomo Geva]

How many rows? Columns? File format?

[Minions]

i have 10039 rows, 64 columns. The file format is csv

[Shlomo Geva]

64 rows? or should that be 10039 rows?

[Minions]

soory sorry my mistake, it is 10039 rows and 64 columns

[Walter Roberson]

30 does not divide evenly into 10039; there are 19 left over. What do you want to do with the final 19?

[Minions]

Thank you for your comment. I stopped my for loop for the last rows, so it didnot count the last rows. Basically, I wrote the programming like below, but the problem is that I get 10011 rows, wheres I should have got around 334 rows after doing the average, it means that I am getting repeated row?

M = csvread('amplitude.csv');
index=1;
   for i=1:30:size(M) -30
    m(index,:)= mean(M(i:i+29,:));
index=index+1;
   end
   csvwrite('test1.csv',m)

[Walter Roberson]

   for i=1:30:size(M,1) - 29

Not -30. end-30+29 -> end-1 . So you only need -29

[Shlomo Geva]

Just use

for i=1:30:floor(size(M)/30)*30

% loop body will execute and skip 30 rows, until there is no more group of exactly 30 rows left.

...

end

Answer 1

[Walter Roberson]

M = your matrix with 10039 rows and 64 columns
N = 30;
whole_blocks = floor(size(M,1)/N);
leftover = size(M,1) - whole_blocks * N;
last_whole = whole_blocks * N;

whole_mean = reshape( mean(reshape(M(1:last_whole, N, [])), 2), whole_blocks, [] );
extra_mean = mean(M(last_whole+1:end, :),1);

overall_mean = [whole_mean; extra_mean];

[Shlomo Geva]

This seems to fail.

e.g . set

M = rand(10039,30);

and the result is:

Error in ... (line 6)

whole_mean = reshape( mean(reshape(M(1:last_whole, N, [])), 2), whole_blocks, [] );

[Walter Roberson]

@Shlomo Geva You are right, my code was buggy.

M = rand(10039,64);

N = 30;
whole_blocks = floor(size(M,1)/N);
leftover = size(M,1) - whole_blocks * N;
last_whole = whole_blocks * N;

whole_mean = reshape(mean(reshape(M(1:last_whole,:), N, [], size(M,2)), 1),[],size(M,2));
extra_mean = mean(M(last_whole+1:end, :),1);

overall_mean = [whole_mean; extra_mean];

size(whole_mean)

ans = 1×2

334 64

size(extra_mean)

ans = 1×2

1 64

size(overall_mean)

ans = 1×2

335 64

[Minions]

thank you

Answer 2

[KALYAN ACHARJYA]

As you mentioned, the size of the data is "10039 columns, 64 rows"

Load the file using readmatrix or load function.

Easiest Way:

cell_data=mat2cell(data,[30,30,4]);
%.........................^ Row data
avg_data=zeros(1,size(cell_data,1));
for i=1:size(cell_data,1)
    avg_data(i)=mean2(cell_data{i});
end
avg_data

Note:

1. If you have the different rows number & divisible by 30 or step size, you can create a 1D row array easily

row_data=30*ones(1,rows/30);

An example rows size is 300, in that case it would be

row_data =
    30    30    30    30    30    30    30    30    30    30

2. As the given rows number is not divisible by 30 exactly (64 case), hence I have created rows vector [30 30 4], you can make it generalize

3. If the all cell elements are equal size in cell data, such case you can easiliy avoid the for loop here by data sequzeeing in cat 3 dimention, afterwards apply the mean on individual planes (3rd Dimension).

Hope it Helps!

[Minions]

thank you for your answer, I tried to run the programming, but i can find error. Also can you please explain how did you do it. I am kinda new in matlab so difficult for me to understand. Thank you in advance

Answer 3

[Shlomo Geva]

%%  load the fille into memory with 
        M = csvread(filename)
%%  loop around M, 30 rows at a time
        for i=1:30:size(M,1)  
            m = mean(M(i:i+29,:)); % compute mean of rows from row i to row i+29
            % here do something with the mean 
        end

[Minions]

hello thank you for your answer but it shows me error for the mean line, i justv need to find the average for each 30 rows

[Shlomo Geva]

Try the following and if that work for you then figure out what is going wrong with your data.

(e.g. check the dimensions of M after reading it from CSV file. Here I make up M, the data matrix:

>> M = [1 2 3 4; 5 6 7 8; 9 10 11 12]

M =

     1     2     3     4
     5     6     7     8
     9    10    11    12

>> mean(M)

ans =

     5     6     7     8

>> 

[Minions]

ok thank you

[Walter Roberson]

        for i=1:30:size(M,1)  

You cannot go right to the end because you take i+29

[Minions]

ok thyank you for your suggestion. I will try this.

[Shlomo Geva]

Yes, Walter Roberson, thanks.

That's an udocumented feature :-)

The number of rows has to be a multiple of 30.

If it is not, then the final few rows have to treated separately. Furthermore, if there is only one row left then the row itself is probably the desirable mean (taking the mean of a vector will give a scalar which not desirable one might think.)

[Minions]

Thank you for explaining it to me

Answer 4

[Shlomo Geva]

Here is a solution to also take care of a number of rows that is not a multiple of your grouping (e.g. 30).

I made up a matrix of random integers (between 1 and 10), having 10 rows and 2 columns.

The code computes the mean of the rows, in groups of 3 rows at a time.

M = randi(10,10,2); % your matrix should go here - read it from csv
disp(M); % show the matrix
N = 3; % your number of vectors in a group that you wish to average (in your case 30)
S = size(M,1); % number of rows in M
for i=1:N:S 
    if i<=S-N 
        % here if we have N rows - safe to average
        m = mean(M(i:i+N-1,:));
    elseif S==i
        % here if only one row left - no need to average
        m = M(i,:);
    else
        % here taking the average of left over rows (more than one).
        m = mean(M(i:end,:));
    end
    disp(m); % replace this by whatever you wish to do with each row 
end