how to find distance matrix using knnclassify in matlab?

I work in image classication and I used two classifiers : svm et knn. the problem is that the knnclassify matlab function allows only return a list of labels test images (class) and I need to get out the distance matrix

    class = knnclassify( TestData , TrainData , TrainLabels);

please help me how can I find the distance matrix ?? and thanks in advance

Answer 1

[Star Strider]

You likely have to use knnsearch to get the distances. See the documentation on Classification Using Nearest Neighbors for details.

[Sahar abdalah]

thank you for your answer, in this example I used a train matrix xapp size 10 * 4 and one for test xtest size 5 * 4 , when I use knnsearch I only get a vector that contains the closest distance but my goal is to have all the distances between samples. is there a way to do it? thanks in advance

load(xapp.mat)
load(xtest.mat)
[n,d]=knnsearch(xapp,xtest,'k',1,'distance','cityblock');
n =

     1
     4
     5
     6
    10

 d =

    0.0441
    0.0762
    0.0100
    0.0354
    0.0189

[Star Strider]

If I understand correctly, the pdist or pdist2 functions may be what you want.

[Sahar abdalah]

thank you, that's exactly the answer

[Star Strider]

My pleasure! I’m glad it worked.