spectrum(x, log="no",span=5, plot=FALSE) similar command in Matlab.

My Matlab code is like following:

del = 1/120; % lenght of sampling interval

subplot(3,1,1)

I= pop(:,2);

I1= I(4801:end)/10000;

plot(t(4801:end),I1,LineWidth=1.5);

xlabel 'Time';

ylabel 'Infectious population, I';

subplot(3,1,2)

S = abs(fft(I1));

L = length(S);

S_oneside= S(1:L/2);

f = (1/del)*(0:L/2-1)/L; %frequency question

S_meg=abs(S_oneside)/(L/2);

P=findpeaks(S_meg);

plot(f,S_meg,LineWidth=1.5);

xlabel 'Frequency';

ylabel 'Spectral density';

subplot(3,1,3)

plot(f,S_meg,LineWidth=1.5)

xlim([0.1 2])

xlabel 'Frequency';

ylabel 'Spectral density';

This gives me the correct graph. But in R for smoothing they use span = 5, how can I smooth the peaks in Matlab?

Answer 1

[William Rose]

@Zakia Hossain,

The curve fiutting toolbox in Matlab includes the function smooth(). By default, smooth applies a moving average filter to the data. You can specify the span, which is the width, in points, of the moving average. Example:

y=sin(2*pi*(0:99)/100)+rand(1,100);  % noisy signal
span=5;
yy=smooth(y,span);  % smoothed signal
plot(0:99,y,'-r',0:99,yy,'-b')
legend('Raw','Smoothed, span=5')

I am not familiar with the smoothing function you used in R, or the meaning of span for the smoothing routine in R. It could be very similar to Matlab's smooth(), as illustrated above.