The matrix is divided in two blocks with an equal variance using the greedy implementation of the partition problem on the covariances between the columns and the total row sums.

1 |

`X` |
numeric array or matrix |

The objective is to find two block with an variance that is as close as possible to the variance of the other block without significantly impacting the performance.

The first step is to calculate the covariances between all columns and the total row sums of the matrix. Using the greedy implementation of the partition problem, we find two groups with variances that are as close as possible to eachother.

numeric binary vector with the partition

Kris Boudt, kris.boudt@vub.ac.be

Edgars Jakobsons, edgars.jakobsons@math.ethz.ch

Steven Vanduffel, steven.vanduffel@vub.ac.be

Kristof Verbeken, kristof.verbeken@vub.ac.be

to test: d<-10; X<-matrix(runif(d^2,max=10),ncol=d); partition<-sample(c(T,F),d,replace=T); print(equalvar(X,partition,"greedy"));print(equalvar(X,partition,"KK"));

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2634669

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.

All documentation is copyright its authors; we didn't write any of that.