# How can I compute the mean value over a time interval?

122 views (last 30 days)
anna bicchi on 23 Apr 2021
Commented: Eric Sofen on 29 Nov 2022
Hi veryone!
I have a matrix of 2 colums (one vector time, one vector values) and i'd like to compute the mean value over a time interval.
How can i do that?
thank you very much in advance

Chad Greene on 23 Apr 2021
Hi Anna, welcome to the forum.
With these types of questions it always helps if you can provide a minimal working example. But you explained the problem well enough that I think I can come up with one.
So you have a matrix M whose columns correspond to time and some dependent variable. Let's say M looks like this:
M = [(1:100)' (1:100)'.^2+randn(100,1)];
Plot the first and second colums to see what kind of data we're looking at:
plot(M(:,1),M(:,2))
xlabel time To get the average over some interval, say between t = 55 and t=65, get the indices of that range:
ind = M(:,1)>=55 & M(:,1)<65;
Then calculate the mean of column 2 over the interval:
mean(M(ind,2))
ans = 3.5481e+03
anna bicchi on 26 Apr 2021
Thank you very very much! :)

Eric Sofen on 7 May 2021
If you have, for example, data once per second for an hour and you want to calculate mean values for each minute, I'd recommend using a timetable to store the data and using the retime function. Something like this
% Synthetic data
t = timetable(seconds(0:3600)',rand(3601,1));
tAvg = retime(t,"minutely","mean");
tAvg.Time.Format = 'm' % Display times in minutes
##### 2 CommentsShowHide 1 older comment
Eric Sofen on 29 Nov 2022
retime(t,"monthly","mean")
The NewTimeStep positional argument lists the possible automatic aggregations around calendar unit. If you wanted to do something custom, you can also specify a new time vector. For example, if you wanted to average into 10-day bins, you might construct a time vector something like:
tvNew = t.Time(1):days(10):t.Time(end)
retime(t,tvNew,"mean")