Convert 10 minute average to hourly

I have some data recorded at 10 minute interval and would like to calculate the hourly average. Some data is filtered out so using reshape is not applicable.

I could write a code to use for and if logicals but I have a lot of data and it would be to "heavy".

For each data I have the value and time (year, month, day, hour, minute (10,20 etc).

Is there a quicker way than using a long for and if code?

Answer 1

[Andrei Bobrov]

Madd = randi([10 20],50,1);
data = [datenum(2013,2,1,0,cumsum(Madd),0), randi(2500,numel(Madd),1)];
% Let your data - two columns -> [date, value]

date1 = datevec(data(:,1));
[a,~,c] = unique(date1(:,1:4),'rows');
out = [a,zeros(size(a,1),2),accumarray(c,data(:,2),[],@mean)];

[sks]

You sir are a gentleman and a scholar. Is it possible to have fun a action/function which is assigned by me. Example instead of finding the mean of the same values, use the atan(x^2/(x+3))? where x is is the values of course.

[sks]

I figured it out, thanks for the original answer.

[Andrei Bobrov]

accumarray(c,data(:,2),[],@(x)atan(x.^2./(x+3)))

Answer 2

[dpb]

Reinsert the missing points w/ NaN and if have Statistics toolbox use NANMEAN(); otherwise it's not difficult to filter them out for the sums and use the RESHAPE "trick" you mentioned...

[sks]

Thanks for the reply, I had considered this option. However I also have to perform another action with the data where NaN cannot be used (would return NaN). Again I could filter out NaN but then the code becomes robust.

[dpb]

That's ok; the accumarray() solution is better, anyway.

I was waiting to hear how the timestamps were stored. Fortunately for you Andrei just made an assumption and went on... :)