CODEWITHSUNDEEP
staticmulti-dimensional-scaling
sam Black
sam Black

Reputation: 33

Relation between Centering matrix and multidimensional scaling

I try to understand the steps of multidimensional scaling, and the the method is based on the centering matrix, I do not understand what's his exact role

Upvotes: 0

Views: 342

Answers (2)

JeeyCi
JeeyCi

Reputation: 597

you need to substitute each x with x-x_mean for each feature/column

p.s. better can use sklearn package in Python - manifold.MDS() from this package - the library will do itself the whole needed transformational operations for your dataset for Multi-Dimensional Scaling

Upvotes: 0

HGB
HGB

Reputation: 1

Multidimensionl scaling computes a set of coordinates in a series of increasing dimensions so you can see which number accounts for major steps in variance reduction, its dimensionality, and which may better be considered randomness, the randomness being spread on all dimensions while a few account for the data. Centering takes a persons data and makes it have the same weight by bringing it to a common variance and standard deviation. How much of this centering makes sense for your data generating process is a matter of some delicacy and has had extensive discussion, particularly with regard to variance. Variance uniformity can be by person or in the case of multiple measures, the measures. Articles on centering by mean is at https://en.wikipedia.org/wiki/Centering_matrix and for MDS is at https://en.wikipedia.org/wiki/Multidimensional_scaling

Upvotes: -1

Related Questions