Remove rows or cols whose elements are all NaN
显示 更早的评论
How can I remove rows or cols whose elements are all NaN ? Withouot any dirty iterations?
For example,
A = [1 1 1 1 1 1 1 1 1 1;
NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN;
1 1 1 1 1 1 1 1 1 1;
NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN;];
should turned into
A = [1 1 1 1 1 1 1 1 1 1;
1 1 1 1 1 1 1 1 1 1];
This is just an example. Actually I have a very big matrix. So I want a solution of this question to work well with my big matrix.
2 个评论
Hi! You can find some clues here: A discussion of the opposite problem on SE. I personally achieved what you were attempting using:
A(~any(~isnan(A), 2),:)=[];
Walter Roberson
2015-9-20
Michal Gajewski commented
works
采纳的回答
Andrei Bobrov
2013-3-25
编辑：Andrei Bobrov
2013-3-25
out = A(all(~isnan(A),2),:); % for nan - rows
out = A(:,all(~isnan(A))); % for nan - columns
6 个评论
Jan
2013-3-25
A can be a row-vector, such that all will operate along the 2nd dimension automatically. Therefore this is safer:
out = A(:, all(~isnan(A), 1));
HUGO LOYA
2014-6-18
Why do you write 1 for columns and 2 for rows?
Jan
2014-7-13
@Hugo: See doc all for details. The numbers specify the dimension to operate on.
Céldor
2015-4-2
Can I obtain a logical matrix I of indices to be held / removed something like
I = something here
and then use assign an altered matrices:
out1 = out1(I);
out2 = out2(I);
% ... etc.
Thanks
gringer45
2018-3-18
This doesn't really do what the question asks for. This selects all the columns or rows with none (zero) NaN values. So, this is answering the question: "Remove rows or cols whose elements have any (at least one) NaN"
verve
2025-1-28
A(:,~all(isnan(A))) % for all nan - columns
A(~all(isnan(A),2),:) % for all nan - rows
更多回答（6 个）
Phillippe
2015-1-14
13 个投票
To remove only ALL-NaN columns, do this instead:
A = A(:,~all(isnan(A)));
4 个评论
Jeffery Devereux
2017-6-20
Thank You!
eda
2019-10-7
thank you so much
Parth Dev Bundela
2022-10-22
Thanks man
osman alper altun
2023-2-14
Thank you!
Use this :
out = A(:,any(~isnan(A))); % for columns
out = A(any(~isnan(A),2),:); %for rows
Azzi Abdelmalek
2013-3-25
A(isnan(A))=[]
6 个评论
Jan
2013-3-25
This does not solve the wanted: "delete rows or cols whose elements are all NaN"
Azzi Abdelmalek
2013-3-25
I am not sure what he want, from his example his rows are all 1 or all nan! In this case the answer solve his problem.
Emanuel-Petre Eni
2016-4-4
Your answer will create an array. he wants to keep the matrix.
Serafeim Zacharopoulos
2020-5-27
To delete rows with all NaN's, maybe try this:
A = A(find(sum(~isnan(A)'))',:)
Walter Roberson
2020-5-28
A(all(isnan(A),2),:) = []; %rows that are all nan
A(:, all(isnan(A),1)) = []; %cols that are all nan
Andrew Sol
2024-11-14
You proposed too simple a solution and you were wrong.
Alfaz Memon
2018-8-20
编辑：Alfaz Memon
2018-8-21
input varibale : data
output variable : row_index( index of rows with all the value as NaN)
row_index( index of rows with all the value as NaN)
column_index( index of columns with all the value as NaN)
if(iscell(data))
x =find(cell2mat((cellfun(@(data) any(isnan(data),2),data,'UniformOutput',false))));
[ia,ib] = ind2sub(size(data),x);
rows_unique = unique(ia);
rows_unique(:,2)=histc(ia,rows_unique);
row_index = rows_unique(find(rows_unique(:,2)==size(data,2)),1);
columns_unique = unique(ib);
columns_unique(:,2)=histc(ib,columns_unique);
column_index = rows_unique(find(columns_unique(:,2)==size(data,1)),1);
else
row_index =find(~any(~isnan(data), 2)); % row with all NaN values
column_index =find(~any(~isnan(data), 1)); %column with all NaN values
end
2 个评论
Walter Roberson
2018-8-20
Seems like a bit of a bother to just remove the rows or columns ?
I notice that you are using cellfun on the data, implying that the data is a cell array; in the original question it was a plain array.
Alfaz Memon
2018-8-21
编辑：Alfaz Memon
2018-8-21
yeah its for cell array. for plain array you can remove cellfun and just simply keep any(isnan(data),2) instead of x =find(cell2mat((cellfun(@(data) any(isnan(data),2),data,'UniformOutput',false))));
this can you work for cell array of different data type.
Also I have updated solution.
Ilham Hardy
2013-3-25
Haven't tried this, but it should works:
A(isnan(A))=[];
1 个评论
Jan
2013-3-25
This does not solve the wanted: "delete rows or cols whose elements are all NaN"
类别
在 帮助中心 和 File Exchange 中查找有关 Operators and Elementary Operations 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
