How to get columns with means that are greater than one, without using loops?

Hi, Its kind of a silly question, but for some reason nothing popped into my mind - I wanna get columns of some matrix where for each column, his mean is greater than one, without using loops. i.e,perform what this code does without the loop:

bigMeters=zeros(1,size(bestMatDiff,2)); %some matrix 
for i_col=1:size(bestMatDiff,2)
    col=bestMatDiff(:,i_col);
    bigMeters(i_col)=(mean(col(col~=0))>1);
end

Thanks, Gal

EDIT: sorry, I forgot to mention that its the mean of the non-zero entries that I want, as can be seen in the code (otherwise its really a dumb question :)

Answer 1

[Andrei Bobrov]

p1 = bestMatDiff;
p1(p1 == 0) = nan;
p2 = nanmean(p1);
out = p1(:,p2 > 1);

or

p1 = bestMatDiff;
t = p1 ~= 0;
p2 = sum(p1)./sum(t);
out = p1(:,p2 > 1);

Answer 2

[Geert]

Hi Gal,

you can find an example in the following code:

% generate random matrix
randomMatrix = 1+randn(10,10);

% specify your threshold (in your case this is 1)
threshold = 1;

% calculate the mean of each column
columnMean = mean(randomMatrix,1);

% if you want to get the "bigMeters" variable from your code, you can add
% the following line of code:
bigMeters = columnMean > threshold;

% the matrix with columns of mean greater than the threshold
newMatrix = randomMatrix(:,columnMean>threshold);

% if you want to now at which column index these columns are, you can find
% them with the "find" command
columnIndices = find(columnMean>threshold);

% % newMatrix can than also be found with the following command:
% newMatrix = randomMatrix(:,columnIndices);

The trick is to use logical indexing, which is done in the line newMatrix = randomMatrix(:,columnMean>threshold);

The find command provides an alternative method.