Dear Sirs, would you please help me and tell me how can I plot the average value of the previous y, while we have thousands (x, y) versus the current X?

(xn, yn), yn=(sigma (i=1) to (i=n) [yn])/n

output(xn, yn=average of previous y and the current y)

We have thousands of dots like (xn, yn), but the number of our data is accidental and it changes by time of sampling. So, there is a necessity to have dynamic code. I think it is not so difficult. Before anything, I am so grateful for your helping. Thank you so much.

[Walter Roberson]

Should it be the average of all of the previous y values, or should it be the average of at most a particular number of previous y values (for example, previous 10) ?

[Hadi Naderiallaf]

Dear Walter, thank you for your consideration. It should be all the previous y values including also the current y. This means that the y in the output should be the average of all the previous y and the available y in the last data. Thanks.

[Hadi Naderiallaf]

Dear Walter, actually, I need a scatter plot in the output including these points: (x1, y1), (x2, average y1 and y2), (x3, average y1, y2 and y3), ........, (xn, average y1, y2, ......, yn). The output should be a graph including thousands of points in the x-y plane. Thank you for your help. I am looking forward to hearing from you.

[Hadi Naderiallaf]

Dear Walter, another point is that all of my data are in text file format. I can also put them in Excel file. Would you please tell me how can I refer the code to the text file or excel file? Thank you so much.

Answer 1

[John D'Errico]

If you just have a vector X and a vector Y, and you want to plot the average of all preceding Y as a function of X, this is trivial and easy to do efficiently.

% some garbage data
N = 10000;
X = rand(1,N);
Y = X + randn(size(X));

[X,sorttags] = sort(X);
Y = Y(sorttags);

% The average of all preceding Y, including the current Y
Ymean = cumsum(Y)./(1:N);

plot(X,Ymean,'.')

WTP?

Or, if you don't want to include the current element in that average...

plot(X(2:end),Ymean(1:end-1),'.')

Or, if you just want the average of the last 10 elements...

Ymean10 = conv(Y,ones(1,10)/10,'valid');
plot(X(10:end),Ymean10,'.')

It would be slightly more difficult if you wanted the mean of only those elements within a specified distance in X less than the current point, because then the number of elements in those means will vary.

[Walter Roberson]

Note: in newer releases, movmean() can be used for moving average.

[Hadi Naderiallaf]

Dear Walter and John, thank you so much for your consideration. The point here is that the number of data is accidental. Thus, in the following code you used N=10000, can not be defined previously before data processing. What should we do with accidental number of points?

[John D'Errico]

N = length(X);

WTP? If you have the data, then you can determine the length of the data.

Answer 2

[Jon]

If you do not want to keep all of the y's in memory and are only interested in the average of all of the y's up to the current sample (N+1) you can do this recursively as: ybar = n/(n+1)*ybar + y/(n+1)

Where ybar is the next value of the average, and n is the total number of point obtained so far. So in this case you just have to keep track of two scalar values, the count n, and the last value of the average, ybar

[Hadi Naderiallaf]

Dear Jonathan, thank you so much for your consideration. The point here is that the number of data is accidental. Thus, in the following code you used N=10000, can not be defined previously before data processing. What should we do with accidental number of points?

[Hadi Naderiallaf]

Dear Jonathan, actually, I need a scatter plot in the output including these points: (x1, y1), (x2, average y1 and y2), (x3, average y1, y2 and y3), ........, (xn, average y1, y2, ......, yn). The output should be a graph including thousands of points in the x-y plane. Thank you for your help. I am looking forward to hearing from you.

[Hadi Naderiallaf]

Dear Jonathan, another point is that all of my data are in text file format. I can also put them in Excel file. Would you please tell me how can I refer the code to the text file or excel file? Thank you so much.

[Jon]

Hello Hadi, In order to work with the data you need to get it into MATLAB arrays, for example a vector with all of the x values and a vector with all of the y values. If the data can be put into an Excel spreadsheet then you can easily import it into numerical arrays using xlsread. The MATLAB documentation for this function will guide you on how to do this. Once you have imported the data, then you can either use John or my method to calculate the average values for y1, (y1,y2), (y1,y2,y3). Regarding your question of what to do if you do not know how many data points (what is the final value of n), you can use the MATLAB length function for this. I think the documentation for this function will help you if you are not familiar with that function.

[Hadi Naderiallaf]

Thank you so much for your nice explanations.

[Hadi Naderiallaf]

Error using xlsread (line 260) Excel Worksheet could not be activated.

[Hadi Naderiallaf]

Dear Jonathan, I wrote the code, but there is the error of: Error using xlsread (line 260) Excel Worksheet could not be activated.

[Hadi Naderiallaf]

Error using xlsread (line 260) Error: The server threw an exception.

[Walter Roberson]

See https://www.mathworks.com/matlabcentral/answers/305976-error-using-xlsread-line-249-excel-worksheet-could-not-be-activated

[Hadi Naderiallaf]

Dear Walter, actually I removed all of my add-in in Excel, especially the Foxit, but there is still the problem.

[Walter Roberson]

Sorry, I do not use Excel on Windows. But see also https://www.mathworks.com/matlabcentral/answers/275384-xlswrite-error-worksheet-could-not-be-activated

[Hadi Naderiallaf]

Thank you so much dear Walter. Now, it is working. The problem was in Excel file the cell with zero values should be removed first. Thank you so much for your help.