Calculate mean values of specific (but dynamic) intervals

Hello everyone,

I want to calculate mean values of specific (but dynamic) intervals across several result files. I know how I get more than one results file into MATLAB and the other basics.

But what I dont know is the following: an interval should be defined/ starting when a value is between -0.05 and 0.05 in a specific column. It is dynamic, because it sometimes lasts for 8 seconds, sometimes for 10 seconds. So the interval should be ended when the value is <-0.05 or >0.05.

Then i want to calculate the mean value of data from another column according to the defined intervals in the other column.

I hope u understand what I want to do.

Thanks in advance

[dpb]

How difficult this may be is dependent upon the characteristics of the signal...if it is relatively noise-free, and the signal doesn't then recross the threshold later, you can simply find the first and last crossings. If there's noise that should be ignored or it does cross later on but outside the triggering interval of interest, then that's got to be factored in as well.

Attaching a sample dataset would probably be best so folks can see what you have to deal with...

[NB]

Thanks for your answer. You're right, so I attached an example with three columns.

The signal is pretty clear but we want to make sure that outlier won't affect the interval. I also had the idea to catch the first and the last crossing, but don't know how to do it.

In the first column is the time (just fyi). The third column represents the data from which the intervals should be defined. In the second column are the values of which I want to calculate the mean-values according to the values in column three.

I coloured them red.

[dpb]

That's a good example of what may be a problem...I presume the initial zero is not of interst, only the intermediate segment?

And, is this the end of the signal or will it eventually cross again, potentially for a real dataset?

[NB]

You're right, the initial zero is not of interest but the intermediate segment. And yes, it will cross again. The original dataset is pretty large with thousands of data lines.

[dpb]

I presumed as much...one last Q? (at least last for now..."one more thing!" :) ) -- is that it will cross again a sure thing and it is only the initial time period within the threshold around zero that is of interest, correct? So one can truncate anything after the transient portion begins once it again at some distant time crosses the threshold? If that is so, then it makes the search logic much simpler.

[NB]

well it will cross it several times (roughly 20-25 times) so there are 20-25 intervals that are of interest and which has to be defined. so unfortunately u cant truncate something

[dpb]

Ah! That's a key piece, then...

Answer 1

[Jos (10584)]

% interval and value are the relevant columns of your data matrix
interval = data(:,3)
value = data(:,2)
% find the sections 
q = ~(interval < -.05 | interval > .05) 
dq = diff([false q false])
ix1 = find(dq==1)
ix2 = find(dq == -1) - 1 
% function to apply to each section
myfun = @(k) mean(value(ix1(k):ix2(k))) 
R = arrayfun(myfun, 1:numel(ix1)) % mean of each section

not tested but it should work :-)

[NB]

Thanks!