How to make a loop to find average of each month

I have a data set of monthly averages and would like to make a loop to get an average of every month, so the average of every January, February, and etc.

Here is my data:

1973 1 144.46 4.68

1973 2 118.77 5.05

1973 3 57.76 7.42

1973 4 103.44 6.01

1973 5 84.11 6.18

1973 6 74.74 6.11

1973 7 64.86 6.46

1973 8 69.95 6.07

1973 9 75.25 5.66

1973 10 91.90 5.64

I can do it the hard way:

a=find(vinddata_mean_monthly(:,2)==1);

jan=vinddata_mean_monthly(a,:);

jan_dir=mean(jan(:,3));

jan_spd=mean(jan(:,4));

..

But if anyone could help my make a loop, help would be greatly appreciated.

Kristine

[Azzi Abdelmalek]

Your question is not clear

[Star Strider]

What is the size of your matrix?

How many complete years do you have?

Answer 1

[dpb]

mndata=zeros(12,2);      % preallocate
for m=1:12               % over each month
  mndata(m,:)=mean(vinddata_mean_monthly(vinddata_mean_monthly(:,2)==m,3:4));
end

ADDENDUM

If the data file is regular in not having any missing months, then one can set the row indices directly and eliminate the logical addressing. Unless the size of the data file is humongous it'll likely not be of any real difference, but still...

for m=1:12
  mx(m,:)=mean(vinddata_mean_monthly(m:12:end,3:4));  % take every 12th row
end

[Kristine]

Perfect, thank you so much!

Answer 2

[Cedric]

Here is one solution if you need monthly means over all years [mean(Jan73,Jan74,..); mean(Feb73,Feb74,..); ..]:

 dir_monthly = accumarray( data(:,2), data(:,3), [12,1], @mean ) ;
 spd_monthly = accumarray( data(:,2), data(:,4), [12,1], @mean ) ;

where data is what you named vinddata_mean_monthly in your question. This works if there are missing data.

[dpb]

Yeah, it's that the input to accumarray must be a vector so couldn't figure out anything that wasn't either a double as you wrote or over accumarray itself that I went w/ the loop, Cedric.

[Cedric]

I often test/profile the loop as well to be honest. Sometimes we think that we do better than a loop with ACCUMARRAY and we end up using CELLFUN/ARRAYFUN/etc to prepare indices because there is no way to get them drectly. And we are mistaken because it looks more elegant and concise but it is not as efficient overall.

Yet, when we are lucky enough to be in a situation where ACCUMARRAY can be used directly, it is difficult to beat.