Is there any matlab function to calculate moving mean square error?

I am looking for a way to calculate mean square error for every 'n' sample in a signal of length N (total number of samples)

[Jonas]

please make clear: do you calculate the least square line once and first and then you want the sliding window of mean error per n sample

OR

do you take a window of n samples, calculate least square line and want to measure the error of that part?

[Kalasagarreddi Kottakota]

Sorry its a mistake, I am looking for to calculate sliding mean square error between two signals.

Answer 1

[Bruno Luong]

Assuming you have 2 signals S1 and S2 in 1 x N arrays:

N = 1000;
S1 = randn(1,N);
S2 = randn(1,N);
n = 10;

dS = S1 - S2;
RMS = sqrt(conv(dS.^2, ones(1,n)/n, 'valid'))

RMS = 1×991

1.1971 1.2146 1.1593 1.0481 1.0444 1.0172 1.0472 1.0975 1.0711 0.9866 1.0295 0.9773 0.9768 1.0108 1.3782 1.3756 1.3565 1.7685 1.8392 1.8426 1.8149 1.8150 1.9150 1.8936 1.6972 1.7878 1.6632 1.1701 1.1023 1.0704

Answer 2

[Mathieu NOE]

hello

I doubt that there is a code for that

try this :

(based on formula) :

% dummy data
n=300;
x=linspace(0,2*pi,n);
f = cos(x) + 0.1*randn(1,n); % values of the model
y = smoothdata(f,'gaussian',30); % actual data

buffer = 10;  % nb of samples in one buffer (buffer size) 
overlap = 9; % overlap expressed in samples

%%%% main loop %%%%
m = length(f);
shift = buffer-overlap;    % nb of samples between 2 contiguous buffers  
for ci=1:fix((m-buffer)/shift +1)
    start_index = 1+(ci-1)*shift;
    stop_index = min(start_index+ buffer-1,m);
    time_index(ci) = round((start_index+stop_index)/2);  % time index expressed as sample unit (dt = 1 in this simulation)
    mse(ci) = my_mse(f(start_index:stop_index) - y(start_index:stop_index));  % 

end
xx = x(time_index); % new x axis 
figure(1),
plot(x,f,xx,mse,'r*');

figure(1),
plot(x,f,'k',x,y,'b',xx,mse,'r');
legend('f data','y data','MSE');



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function x_mse = my_mse(x)

    x_mse = mean(x.^2);
    
end