Moving mean of a vector with unsorted values

Hello everyone,

I have a vector of measured angle values (1x61 double), which was measured in this direction: 15:-1:-15 and then in the reverse direction as -14.5:1:14.5 degree. So the resulting vector is [15 14 13 ... -13 -14 -15 -14.5 -13.5 -12.5 ... 12.5 13.5 14.5]. Then I have corresponding measured Cd values for the angles, also 1x61 double.

I tried to do a moving average

Cdormov = movmean(Cd_original, 5);

Unrecognized function or variable 'Cd_original'.

When I plot the results, I can see that Matlab calculated the moving average for the Cd values as for two vectors. However, I need Matlab to treat the angle and Cd data as one X and one Y vector and calculate only one moving mean curve. How can I do this? Thank you.

The plot code:

figure(4) 
hold on; grid on; grid minor;
plot(angle,Cd_original,'black-+');
plot(angle,Cdormov,'red-o');
xlabel('angle (deg)'); ylabel('Cd (-)');
legend('original data','moving mean');
title(['Cd ' name])

The plot:

Answer 1

[Voss]

% specified angle vector
angle = [15:-1:-15 -14.5:1:14.5];

% example Cd_original vector
Na = numel(angle);
Cd_original = sin(angle/4)+0.5*round((0:Na-1)/Na);

% sort angle, and then calculate the moving mean on
% Cd_original reordered according to angle_sorted
[angle_sorted,idx] = sort(angle);
Cdormov = movmean(Cd_original(idx), 5);

% when you plot, plot Cdormov against angle_sorted
name = 'V_o21_23proc';
figure(4) 
hold on; grid on; grid minor;
plot(angle,Cd_original,'black-+');
plot(angle_sorted,Cdormov,'red-o')
xlabel('angle (deg)'); ylabel('Cd (-)');
legend('original data','moving mean');
title(['Cd ' name])

[Voss]

Compare the results of unique-ing vs merely sorting, in a case with lots of duplicate values in the angle vector:

% angle vector
angle = [5:-0.5:-5 -4.5:0.5:4.5];

% lots of repeated values
numel(angle), numel(unique(angle))

ans = 40

ans = 21

% example Cd_original vector
Na = numel(angle);
Cd_original = sin(angle)+0.5*round((0:Na-1)/Na);

% use unique and sort and see what happens
[angle_unique,idx_unique] = unique(angle);
Cdormov_unique = movmean(Cd_original(idx_unique), 5);

[angle_sort,idx_sort] = sort(angle);
Cdormov_sort = movmean(Cd_original(idx_sort), 5);

figure(4) 
tiledlayout(2,1)
nexttile()
hold on; grid on; grid minor;
plot(angle,Cd_original,'black-+');
plot(angle_unique,Cdormov_unique,'red-o')
xlabel('angle (deg)'); ylabel('Cd (-)');
legend('original data','moving mean');
title('with unique()')
nexttile()
hold on; grid on; grid minor;
plot(angle,Cd_original,'black-+');
plot(angle_sort,Cdormov_sort,'red-o')
xlabel('angle (deg)'); ylabel('Cd (-)');
legend('original data','moving mean');
title('with sort()')

●

Answer 2

[Stephen23]

You will need to interlace your data points before calling MOVMEAN. For example:

xL = +15.0:-1:-15.0;
xR = -14.5:+1:+14.5;
yL = sin(xL/4);
yR = sin(xR/4)+0.5;
yU = [yL,yR];
[xU,idx] = unique([xL,xR]);
yU = yU(idx);
yM = movmean(yU,5);


plot(xL,yL,'-+k', xR,yR,'-xk', xU,yM,'-*r')

●