How can I calculate a weighted mean in Matlab?

I have a 180-by-360 matrix of (surface temperature) values and I want to calculate a weighted average of all values given in this matrix. However, I need to weight these values with respect to latitude. Is there a way to calculate a weighted mean in Matlab? Please help me.

Thanks in advance for any help!

- Jan-Erik

[shelley]

Normally taking dim 1 mean of some, say 3x5 matrix A, can be seen as w=[1/3 1/3 1/3], and w*A returns you the answer.

Now if we want to take dim 1 weighted mean of A, we can set our weight in w = [1/6 2/3 1/6], then take w*A, which should return you a 1x5 row vector.

Answer 1

[Matt J]

If A is your 180x360 matrix and W is a 180x1 vector of weights for the latitudes, do

 weighted_mean = mean(W.'*A,2);

[Youssef Khmou]

hi, i think the question is how to generate the weights ( normalized or not) , do you have an idea on how to do so using Gaussian kernel?

[Jan-Erik]

Thank you for your help!

Matt J - your answer seems not to work. Is there a typo? Here's the command lines: A = rand(180,360).*100; w = cos(linspace(-pi/2,pi/2,180)); weighted_mean = mean(w.'*A,2) ??? Error using ==> mtimes Inner matrix dimensions must agree.

Youssef - I'm not sure what you mean with Gaussian kernel.

[Matt J]

I specified my assumption that w is 180x1 (i.e., a column vector). Never mind, though. You can just omit the transpose if you want to maintain it in row form.

 weighted_mean = mean((w/sum(w))*A,2)

I've also normalized w here so that the weights sum to one.

[Jan-Erik]

Thanks!

[Jason Garcia]

weighted_mean = mean(bsxfun(@times,A,w'),2)

[Matt J]

weighted_mean = mean(bsxfun(@times,A,w'),2)

If that's what the OP intended, it would be more efficient to do

 weighted_mean=w'.*mean(A,2);

Answer 2

[John Hock]

Hi Everyone

I just want to ask one thing that if i have 2d array 2*4 and want to get weighted average with vector of 1*4

i want first row of A get multiplied with B and give weighted average as an output similarly for next row.

Please help

[Image Analyst]

Try

B = repmat(B, [2, 1]) % Make B 2*4
weightedMeans = sum(A.*B, 1); % Get weighted means within a row going across columns.

[stelios loizidis]

Hello,

I have the following questuin: I have a matrix e.g

[5 3 3 1
 3 4 5 2
 5 0 0 0
 3 4 5 2]

For this matrix I have to make a weighted average for each row. That is [5+3+(4/5)*(3+1)] /4=16/5. Also, if there is only one number and the zero (row 3) the weighted average should be [5+0+0+0] / 1 =5 and the end we choose the largest weight average. How does this work?

Answer 3

[Micke Malmström]

See this function https://se.mathworks.com/matlabcentral/fileexchange/14416-wmean?s_tid=srchtitle