Multidimensional Scaling for Big Data
MDS is a statistic tool for reduction of dimensionality, using as input a distance
matrix of dimensions n × n. When n is large, classical algorithms suffer from
computational problems and MDS configuration can not be obtained.
With this package, we address these problems by means of three algorithms:
- Divide-and-conquer MDS proposed by Delicado P. and C. Pachón-García (2021)
- Interpolation MDS, also proposed by Delicado P. and C. Pachón-García (2021)
<2007.11919>, which uses Gower's interpolation formula as described in
Gower, J. C. and D. J. Hand (1995).
- Fast MDS, which is an implementation of the algorithm proposed by
Yang, T., J. Liu, L. McMillan, and W. Wang (2006).
The main idea of these algorithms is based on partitioning the data set into small
pieces, where classical methods can work. In order to align all the solutions,
Procrustes formula is used as described in Borg, I. and P. Groenen (2005).2007.11919>2007.11919>