How to remove corresponding rows of a second array in the case of NaNs in the first array?

I have a double array called mean with e.g. 11 rows and 121 columns, where row 3, 6, and 7 are NaNs. I want to remove these NaNs, and also remove the corresponding rows of another array called clst.

I tried to remove the NaNs with the following code:

mean(all(isnan(mean),2),:) = [];

But i would like to create a for loop where for each row of 'mean' it would detect the NaNs, and remove these rows and the corresponding rows of 'clst'. Would someone be able to help me?

[Stephen23]

Do NOT name your variable mean, because that is the name of a very important inbuilt function.

[Mathieu NOE]

hi

it's not good pratice to give matlab native function names to variables

here we don't know by sure if mean is your variable or the matlab mean function

quite confusing

Answer 1

[Stephen23]

Where M is your matrix (do not use the name mean):

idx = all(isnan(M),2);
M(idx,:) = [];
clst(idx,:) = [];

Why do you want to use a loop?

[Aafke Kerkhof]

Because I am completely new to MATLAB and I thought that would be the only possibility, but this worked like a charm thanks!

Answer 2

[Bjorn Gustavsson]

Something like this might be preferable to a straightforward loop:

x1 = randn(5,11);
x2 = x1;
x1(x1>1.5) = nan; 
idxNaN = any(isnan(x1),2); % Or if you have entire rows with nans, then use all
x2(idxNaN,:) = [];
x1(idxNaN,:) = []; % or whatever "saving-operation" you want to do on those rows

Also worth repeating (even though it is nagging and grating): avoid using variable-names like mean that shadows the built-in commands, nothing but problems will come out of that. Instead use something more descriptive, like x_avg or x_mean. Then you instantly also see what you have the mean of.

HTH

Answer 3

[Mathieu NOE]

hello again

a for loop is maybe overkill , but as it's what your are asking for , this is my suggestion :

i prefered the names A and B for your corresponding mean (ugh!) and clst arrays

% dummy data
A = rand(11,121);
A([3;6;7],:) = NaN;

B= rand(11,12);
B([3;6;7],:) = NaN;

% main code 
[m,n] = size(A);
k = 0;
A2 = [];
B2 = [];

for ci = 1:m
    tmp = A(ci,:);
    if ~any(isnan(tmp))
        k = k +1;
        A2(k,:) = tmp;
        B2(k,:) = B(ci,:);
    end
end