Compute distances between the rows of a matrix.
Syntax
d = pdist(m)
d = pdist(m, metric)
d = pdist(m, 'minkowski', p)
Inputs
- m
- The matrix on which to operate.
- Type: double
- Dimension: matrix
- metric
- The distance function to compute.
- The available options are as follows:
- 'euclidian' (default)
- Euclidean distance.
- 'sqeuclidian'
- Square of the Euclidean distance.
- 'seuclidian'
- Standardized Euclidean distance, the Euclidean distance
normalized by the sample variance.
- 'mahalanobis'
- Mahalanobis distance.
- 'cityblock'
- City Block distance.
- 'minkowski'
- Minkowski distance, also known as the p-norm.
- 'cosine'
- One minus the included angle between two vectors.
- 'correlation'
- One minus sample correlation between two vectors.
- 'spearman'
- One minus sample Spearman's rank correlation between two vectors.
- 'hamming'
- Hamming distance.
- 'jaccard'
- One minus the Jaccard distance.
- 'chebychev'
- Chebychev distance, the maximum coordinate difference.
- Type: string
- p
- The parameter in the Minkowski distance (default: 2).
- Type: integer
- Dimension: scalar
Outputs
- d
- The distance values.
- Dimension: vector
Example
m = [1,2,3;4,6,8;5,10,15;16,26,36];
d = pdist(m, 'cityblock')
p = [Matrix] 1 x 6
12 24 72 12 60 48
Comments
If m contains r rows then d is a row vector with
r*(r-1)/2 elements. The first r-1 elements are the distances between the first row
and the remaining r-1 rows. The next r-2 elements are the distances between the second
row and the remaining r-2 rows, and so on.
