Hi, please I want to calculate the distance between this tow matrix using pdist2

Answer 1

[John BG]

Olfa

    p=[26 255 255 255 0 255 255]
    s=[255 255 255 255 0 255 255]

 % vector distance from p to s

    s-p
    =
     229     0     0     0     0     0     0

the euclidean distance is the magnitude or norm of the vector distance

   s_p_dist_euclid=pdist2(s,p,'euclidean')
   =
     229

the square euclidean distance is the area of the square with side the previous euclidean distance

s_p_dist_sqeuclid=pdist2(s,p,'squaredeuclidean')
 =
       52441

229*229
 =
       52441

citiblock distance

   s_p_dist_cityblock=pdist2(s,p,'cityblock')
  s_p_dist_cityblock =
     229

minkowski default coeff is 2

>> s_p_dist_mink=pdist2(s,p,'minkowski')
s_p_dist_mink =
   229

   s_p_dist_mink=pdist2(s,p,'minkowski',1.5)
  s_p_dist_mink =
       2.289999999999999e+02

Cheby distance: maximum coordinate difference

   s_p_dist_chevy=pdist2(s,p,'chebychev')
  s_p_dist_chevy =
     229

cosine distance: 1-cos(alpha) alpha is the angle between vectors s and p

   s_p_dist_cos=pdist2(s,p,'cosine')
  s_p_dist_cos =
     0.069480529510180

correlation distance: 1-sample correlation, it means, the opposite of the correlation

   s_p_dist_corr=pdist2(s,p,'correlation')
  s_p_dist_corr =
     0.307354858007336

the correlation distance of of same vectors would be null because it's max correlated

s_p_dist_corr=pdist2(p,p,'correlation')
 =
     2.220446049250313e-16

hamming distance: the percentage of coordinates that differ

s_p_dist_hamming=pdist2(s,p,'hamming')
 =
      0.142857142857143

only one coordinate different:

   1/7 = 0.142857142857143

Jaccard distance: percentage of nonzero coordinates that differ

s_p_dist_jaccard=pdist2(s,p,'jaccard')
 =
   0.166666666666667

you also define a custom distance function

D = pdist2(X,Y,@distfun)

the custom distance function has to have this header

function D2 = distfun(ZI, ZJ)

If you find this answer of any help solving your question, please click on the thumbs-up vote link, or mark it as accepted answer

thanks in advance

John

[e-mail]

Answer 2

[Walter Roberson]

You did calculate the distance; it is stored in val .

You are seeing a warning, not an error. To get rid of the warning,

pdist2(double(p), double(s), 'euclidean')

[Walter Roberson]

You will find that either your p or s or both are type uint8 and that is what is triggering the warning.