mean of all these variables over latitude x longitude x time that is (5 x 5 x 2)

I have eight variables loaded with data (latitude,longitude,time,months) as (5,5,2,65). I want to calculate the mean of all these variables over latitude x longitude x time that is (5 x 5 x 2). My output should have the dimensions of (65,8), that is mean values of 8 variables for 65 months.

I would be highly obliged to get fews lines of matlab code to do it.

regards,

Devendra

[Stephen23]

"I have eight variables loaded with data (latitude,longitude,time,months) as (5,5,2,65)."

Rather than eight separate variables, you should store that data in one array in the first place.

For example, a cell array:

C = arrayfun(@(m)rand(5,5,2,65),1:8,'uni',0); % fake data

Then your task is much easier and generally more efficient, e.g.:

M = permute(mean(cat(5,C{:}),1:3),[4,5,1,2,3])

M = 65×8

0.5092 0.5678 0.5696 0.5255 0.4764 0.4271 0.5042 0.5805 0.5234 0.5244 0.5329 0.4640 0.5409 0.5249 0.4913 0.5198 0.4914 0.4635 0.4210 0.4762 0.5566 0.4954 0.4357 0.5009 0.4752 0.5219 0.5962 0.4758 0.4505 0.5813 0.4866 0.4920 0.5215 0.5697 0.5789 0.5397 0.4757 0.5054 0.5671 0.4615 0.5384 0.4807 0.4889 0.4578 0.5242 0.4778 0.5621 0.5629 0.3929 0.5005 0.5730 0.5106 0.4421 0.5782 0.4736 0.5686 0.5001 0.5216 0.4456 0.5744 0.5294 0.4817 0.5263 0.5271 0.4821 0.4879 0.4886 0.4765 0.5305 0.4724 0.5353 0.5123 0.5306 0.4516 0.4656 0.5014 0.4779 0.4436 0.4413 0.5089

M = reshape(mean(cat(5,C{:}),1:3),[],numel(C)) % See Arya Chandan Reddy's answer

M = 65×8

0.5092 0.5678 0.5696 0.5255 0.4764 0.4271 0.5042 0.5805 0.5234 0.5244 0.5329 0.4640 0.5409 0.5249 0.4913 0.5198 0.4914 0.4635 0.4210 0.4762 0.5566 0.4954 0.4357 0.5009 0.4752 0.5219 0.5962 0.4758 0.4505 0.5813 0.4866 0.4920 0.5215 0.5697 0.5789 0.5397 0.4757 0.5054 0.5671 0.4615 0.5384 0.4807 0.4889 0.4578 0.5242 0.4778 0.5621 0.5629 0.3929 0.5005 0.5730 0.5106 0.4421 0.5782 0.4736 0.5686 0.5001 0.5216 0.4456 0.5744 0.5294 0.4817 0.5263 0.5271 0.4821 0.4879 0.4886 0.4765 0.5305 0.4724 0.5353 0.5123 0.5306 0.4516 0.4656 0.5014 0.4779 0.4436 0.4413 0.5089

[Devendra]

How to aviod NaN numvers in data set?

Thanks.

Devendra

[Stephen23]

"How to aviod NaN numvers in data set?"

Use the OMITNAN flag.

Answer 1

[Prannoy]

To calculate the mean of the variables over the dimensions (latitude x longitude x time) and obtain an output with dimensions (65, 8), you can use the mean function in MATLAB along with appropriate reshaping and transpose operations. Here's an example code snippet that demonstrates this:

% Calculate the mean over latitude x longitude x time
mean_values = mean(reshape(variable1, [], 5, 5, 2), [2, 3]);

Unrecognized function or variable 'variable1'.

% Reshape the mean values to have dimensions (65, 8)
mean_values = reshape(mean_values, 65, []);

% Concatenate the mean values with the other variables
output = [mean_values, variable2, variable3, variable4];

% Display the output matrix
disp(output);

In this code, variable1 represents the first variable for which you want to calculate the mean. You can repeat the process for the other variables by replacing variable1 with the corresponding variable names (variable2, variable3, etc.).

Answer 2

[Arya Chandan Reddy]

Hi Devendra, I understand that you are trying to calculate mean of your data (5x5x2) over 65 months for each of the 8 variables.

Assuming that you data is in the format 5x5x2x65x8

Here is the code:

x = rand([5 5 2 65 8]);
m  = mean(x , [1 2 3]);
m = reshape(m , [65 8]);

Refer to the documentation of mean and reshape for better understanding.

Hope it helps

[Devendra]

data format for each variable is (5,5,2,65) not (5,5,2,65,8) as presumed by you and there are 8 variables

[Stephen23]

+1 tidy solution, neat use of RESHAPE. Using one array is good advice,

"data format for each variable is (5,5,2,65) not (5,5,2,65,8) as presumed by you and there are 8 variables"

You can simply use CAT to join them together:

x = cat(5,A1,A2,..,A8);

Even better is to not have eight separate variables in the first place.

Answer 3

[DGM]

If your data is a bunch of individual numeric arrays:

% some fake data 
x1 = rand(5,5,2,65);
x2 = rand(5,5,2,65);
x3 = rand(5,5,2,65);
x4 = rand(5,5,2,65);
x5 = rand(5,5,2,65);
x6 = rand(5,5,2,65);
x7 = rand(5,5,2,65);
x8 = rand(5,5,2,65);

allmeans = [squeeze(mean(x1,1:3)) ...
            squeeze(mean(x2,1:3)) ...
            squeeze(mean(x3,1:3)) ...
            squeeze(mean(x4,1:3)) ...
            squeeze(mean(x5,1:3)) ...
            squeeze(mean(x6,1:3)) ...
            squeeze(mean(x7,1:3)) ...
            squeeze(mean(x8,1:3))];

Alternatively,

% or maybe use a cell array
C = {x1 x2 x3 x4 x5 x6 x7 x8};
allmeans = cellfun(@(x) squeeze(mean(x,1:3)),C,'uniform',false);
allmeans = cell2mat(allmeans);

[Devendra]

data contains NaN values. How to avoid NaN values.

[DGM]

If you want to simply ignore the NaNs, you can use the 'omitnan' flag in the calls to mean().

mu = mean([1 2 3 4 NaN],'omitnan')

mu = 2.5000

Answer 4

[Peter Perkins]

It's very likely that you should be using a timetable.

You have "eight variables loaded with data (latitude,longitude,time,months) as (5,5,2,65)". I think you mean that you have 8 5x5x2x65 arrays. Here's what I recommend:

create the lat/lon/time/month values as 5/5/2/65-element vectors, respectively. Use ndgrid to expand those out to four 5x5x2x65 matrices, where each one of them has a lot of repeated lat/lon/time/month values. Turn those four "coordinate" arrays and your eight data arrays into columns using (:), and put those into a (5*5*2*65)-by-(4+8) table. Now call groupsummary to compute monthy means, or month-by-time means, or lat-by-lon means.

Depending on what timestamps you have, you may want to look at using datetimes, or durations, and putting your data in a timetable. Not enough info to go on.

Answer 5

[Cris LaPierre]

See this answer: https://www.mathworks.com/matlabcentral/answers/1997658-how-do-i-compute-mean-values-of-nine-variables-over-lat-x-lon-x-time?s_tid=mlc_ans_email_view#comment_2822438