Nearest neighbour for one dataset

Question

AB 2017-9-1

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/354890-nearest-neighbour-for-one-dataset

评论： Walter Roberson 2017-9-1

Hello,

I have one dataset consiting of 3d points. I want to find out within that dataset, which point is closest to which in 3d? Eg; [1 2 3; 3 4 5;6 7 8;8 9 10]... And what are some neat tricks to visualize such results?

Thanks!

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2017-9-1

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/354890-nearest-neighbour-for-one-dataset#answer_280000

https://www.mathworks.com/help/stats/knnsearch.html

To visualize, you could construct a directed graph using digraph() and plot it with coordinates https://www.mathworks.com/help/matlab/ref/graph.plot.html#buzeikk

2 个评论
显示无隐藏无

AB 2017-9-1

Hello Walter,

I looked at it. Thanks. I have one question though. The knnsearch works for 2 matrices X and Y. If I have the same dataset as X and Y, will it not count the nearest neighbour for a point in X as the same point in Y. How can I get rid of this?

Thanks!

Walter Roberson 2017-9-1

在 MATLAB Online 中打开

%calculate distances of each point to each other point
dists = squareform( pdist(X) );
%set diagonal to inf so point is not its own closest neighbor
dist(1:size(dists,1)+1:end) = inf;
%find index of closest point
[~, minidx] = min(dists, [], 2);

Now for the K'th point, X(K,:), minidx(K) is the row index of its closest neighbour.

For large number of points, the above might not be as efficient as what knnsearch does, but for modest number of points it is probably more efficient.

请先登录，再进行评论。