Removing rows that are not unique from an array?

Is there an easy way to remove ALL rows that are NOT unique? For example, how would I get B from A?
A = [1 2; 1 3; 1 4; 1 2; 1 5];
B = [1 3; 1 4; 1 5];
I could do this in a loop, but there seems like there must be a more elegant way. I've looked at various applications in the forum using the unique() function, but a solution is not obvious to me.
Thanks!

 采纳的回答

This works:
A = [1 2; 1 3; 1 4; 1 2; 1 5];
[~,ia,ic] = unique(A, 'rows'); % Unique Elements
v = accumarray(ic, 1); % Tally Occurrences Of Rows
B = A(ia(v==1),:) % Keep Rows That Only Appear Once
B =
1 3
1 4
1 5

5 个评论

+1. It took me a while to understand, how to treat this as Run Length Encoding problem.
Thank you, Star Strider. This looks perfect. I was not aware of the accumarray() function...until now!

请先登录,再进行评论。

更多回答(1 个)

As = sortrows(A);
k = find([true; any(diff(As, 1, 1), 2); true]);
B = As(k(diff(k) == 1), :);
And if the original order is wanted:
[As, idx1] = sortrows(A);
k = find([true; any(diff(As, 1, 1), 2); true]);
idx2 = k(diff(k) == 1);
B = A(idx1(idx2), :);
For A = randi([1, 20], 1e5, 4) the first method is 15% faster than the unique/accumarray method.

3 个评论

Thank you, Jan! I will try this method also.
This is asked so often it should be in the FAQ. But before I do, I'd like to have a solution to the other case people ask a lot about, and that is where people want to keep the first instance of the duplicate row (along with unique rows), rather than toss out all rows that are members of duplicates. Another case might be to keep only the duplicate rows.
Isn't "keeping the first instance" solved by
unique(A, 'rows', 'first')
And for keeping the multiple occuring rows only use:
As = sortrows(A);
k = find([true; any(diff(As, 1, 1), 2); true]);
B = As(k(diff(k) > 1), :);
% ^ instead of ==

请先登录,再进行评论。

类别

帮助中心File Exchange 中查找有关 Structures 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by