INTRDIM

Estimate Intrinsic dimension from dissimilarity data

    K = INTRDIM(D)
    K = D*INTRDIM

Input
D	NxN Dissimilarity dataset

Output
K	Estimated intrinsic dimension

Description

It is assumed that the data is generated from a normal distribution in  K dimensions and that the Euclidean distance measure has been used to  compute D.  Do NOT supply squared distances! They will definitely generate the wrong  result.

Example(s)

 A = GENDATB(100);
 D = SQRT(DISTM(A));
 K = INTRDIM(D);

 A = GENDATD(100,10);
 D = SQRT(DISTM(A*SCALEM(A,'VARIANCE')));
 K = INTRDIM(D);