Calculate average over each specific interval

Hi,

I'm trying to calculate the mean every 5 seconds of data till the end.

so x= (1,63369) % my data in Newton Meter which is about 30.9419 seconds long

t = 5*2048 % 5 seconds interval.

I'd like to have your help in (1- how to make it calculate the mean in increaments from start till the end 2- how to deal with the last posrtion of the data that shorter than 5 seconds)

any help would be appreciate it. Thanks in advance.

[Image Analyst]

Please attach your time and signal data (I'm guessing it's t and x) in a .mat file with the paperclip icon

save('answers.mat', 't', 'x');

so we can try some things.

TUTORIAL: How to ask a question (on Answers) and get a fast answer

[Saad Alqahtani]

Definitely. Please see attachment. Thanks!

Answer 1

[dpb]

2) first -- how to deal with the last posrtion of the data that shorter than 5 seconds"

How do you WANT to deal with it? You can either ignore it and have fix(numel/5) samples or average the values that do have in the last segment -- it's your choice.

1) The classic MATLAB way is to reshape() and use mean() --

a) Ignore last partial segment --

N=5;                                                % number elements to average over
mnX=mean(reshape(x(1:N*fix(numel(x)/N)),N,[]));     % average over N elements ignoring end values

b) Deal with odd end

x=[x nan(1,N*ceil(numel(x)/N)-numel(x))];           % augment vector to make even
mnX=mean(reshape(x,N,[]),1,'omitnan');              % average over N elements ignoring NaN elements

[Saad Alqahtani]

Answer ( a ) works really well . However, answer (b) gave me this error "Error using reshape

Product of known dimensions, 10240, not divisible into total number of elements, 63369."

I'd would appreciate it if you can help me with another solution. Otherwise, I'm satisfied with answer (a). Thank you so much.

[dpb]

Must've made a typo somewhere; works for me---

>> x=rand(1,63369);
>> numel(x)
ans =
       63369
>> N
N =
     5
>> ceil(numel(x)/5)
ans =
       12674
>> N*ceil(numel(x)/N)
ans =
       63370
>> ans/N
ans =
       12674
>> x=[x nan(1,N*ceil(numel(x)/N)-numel(x))];
>> numel(x)
ans =
       63370
>> mnX=mean(reshape(x(1:5*fix(numel(x)/5)),5,[]));
>> whos mnX
  Name      Size                Bytes  Class     Attributes

  mnX       1x12674            101392  double              

>> 

[Saad Alqahtani]

(b) works now as well. It looks like I had some typo somewhere. Thank you so much. I really appreciate your help!