calculate monthly averages of time series

I have a dataset which contains measurements taken at discrete intervals for the past ~ 40 years. I would like to calculate the monthly mean values from the measurements. Consider the following example:

DateTime=datestr(datenum('1970-01-01','yyyy-mm-dd'):1:...
    datenum('2011-12-31','yyyy-mm-dd'),...
    'yyyy-mm-dd');
DateTime=cellstr(DateTime);
DateVec = datevec(DateTime);
dt = [1,4,32,232,101,44,23,1,87,843,456];
DateVec(dt,:) = [];

Data = rand(1,length(DateVec(:,1)));

Here, I have excluded some of the Dates in order to specify that the intervals of measurements were not continuous, obviously the measurements are far more spread out than shown above. So, what I would like to do is to use the date vector to specify each year and then calculate the mean of each month for a given year so that at the end I would have 12 measurements for each year.

Answer 1

[Andrei Bobrov]

[a,b,b] = unique(DateVec(:,1:2),'rows');
out = [a,accumarray(b,Data,[],@mean)]; %EDIT row

or

[a1,b1,b1] = unique(DateVec(:,1));
[a2,b2,b2] = unique(DateVec(:,2));
out = [nan,a2';a1,accumarray([b1,b2],Data,[],@mean)]; % EDIT row

[Richard]

many thanks. From this, to calculate the average from each month, not the accumulation I could write:

NewVal = cell2mat(arrayfun(@(x)length(find(b == x)),unique(b),'un',0));
out(:,3) = out(:,3)./NewVal;

Is this correct?

[Andrei Bobrov]

Sorry! I edited my answer. Please read about MATLAB's function accumarray.

[Benju Baniya]

Hi, i am new to matlab and trying to work with timeseries data. What does a, b, b mean here? My data looks like as shown below. How do I calculate GPP based on month so that I get 12 values for each year? Thanks in advance.

YEAR TIMESTAMP DoY NEE GPP

2013 201301010030 1 -0.0064 0.0025

2013 201301010100 2 - 0.0076 0.0026

2014 201401010030 1 0.0983 -0.0076

2014 201301010100 2 0.0075 - 0.0089