Calculate mean of an array in time windows.

Hi

i have EEG data recorded at 1200hz from 5 channels, for 10 repetitions, and i need to calculate mean of each 1s signal in 10 time windows (each of 100ms) as a features. Is thi the right way to do.? Although the features i'm getting isn't right.

for repetition = 1:size(EEG,3) % loop through repetitions

featCnt = 1; % feature counter for generic choice of features or channels

chanCnt = 1;

for channel = 1:length(channels)

for window = 1:10

% calculate mean for current repetition and channel

features(repetition,featCnt) = mean(EEG(repetition)+(1200/10)*(window-1):(EEG(repetition)+(1200/10)*(window+1)-1));

featLabels{featCnt} = sprintf('%s_Mean_%i',chanLabels{chanCnt},window);

featCnt = featCnt + 1; % Count up, for generic choice of feature or channels

end

chanCnt = chanCnt + 1;

end

end

Answer 1

[Athul Prakash]

I'm not sure how you have arranged your data within the matrix 'EEG' - which dimensions correspond to which attributes?

I noticed that in the following line:

features(repetition,featCnt) = mean(EEG(repetition)+(1200/10)*(window-1):(EEG(repetition)+(1200/10)*(window+1)-1));

you are indexing into EEG using just one variable 'repetition'. But you seem to be looping through dimension-3 of EEG with the same variable - this would produce the wrong result.

I would recommend understanding Matlab's indexing thoroughly before you get working on matrix math such as this. Have a look at this doc:

https://www.mathworks.com/company/newsletters/articles/matrix-indexing-in-matlab.html

If you need to brush up on or practice working with loops, indexing etc, Maltab OnRamp is a good place to start.

https://www.mathworks.com/learn/tutorials/matlab-onramp.html