Expected share of instances in slice (independent dimensions).
Expected share of instances in marginal restriction (reference dimension). Added with respect to the original paper to loose the dependence of beta from alpha.
Expected share of instances in slice (independent dimensions).
Expected share of instances in marginal restriction (reference dimension).
Compute the contrast of a subspace
Compute the contrast of a subspace
The indexes from the original data ordered by the rank of the points
The dimensions in the subspace, each value should be smaller than the number of arrays in m
The contrast of the subspace (value between 0 and 1)
A data set (row oriented)
Compute the contrast of a subspace // This is a version where alpha is always choosen at random between 0.1 and 0.9
Compute the contrast of a subspace // This is a version where alpha is always choosen at random between 0.1 and 0.9
The indexes from the original data ordered by the rank of the points
The dimensions in the subspace, each value should be smaller than the number of arrays in m
The contrast of the subspace (value between 0 and 1)
Compute the pairwise contrast matrix for a given data set Note: This matrix is symmetric
A data set (row oriented)
Compute the deviation of a subspace with respect to a particular dimension
Compute the deviation of a subspace with respect to a particular dimension
The indexes from the original data ordered by the rank of the points
The dimensions in the subspace, each value should be smaller than the number of arrays in m
The reference dimensions, should be contained in dimensions
A 2-D Array contains the contrast for each pairwise dimension
Compute the deviation of a subspace with respect to a particular dimension // This is a version where alpha is always choosen at random between 0.1 and 0.9
Compute the deviation of a subspace with respect to a particular dimension // This is a version where alpha is always choosen at random between 0.1 and 0.9
The indexes from the original data ordered by the rank of the points
The dimensions in the subspace, each value should be smaller than the number of arrays in m
The reference dimensions, should be contained in dimensions
A 2-D Array contains the contrast for each pairwise dimension
Compute the pairwise deviation matrix for a given data set Note: This matrix is asymmetric
Compute the pairwise deviation matrix for a given data set Note: This matrix is asymmetric
The indexes from the original data ordered by the rank of the points
A 2-D Array contains the deviation for each pairwise dimension
Compute a statistical test based on Mann-Whitney U test using a reference vector (the indices of a dimension ordered by the rank) and a set of Int that correspond to the intersection of the position of the element in the slices in the other dimensions.
Compute a statistical test based on Mann-Whitney U test using a reference vector (the indices of a dimension ordered by the rank) and a set of Int that correspond to the intersection of the position of the element in the slices in the other dimensions.
The original position of the elements of a reference dimension ordered by their rank
An array of Boolean where true means the value is part of the slice
The Mann-Whitney statistic
Simply like MWP but does not correct ties (but adjust ranks still)
Expected share of instances in slice (independent dimensions).
Expected share of instances in marginal restriction (reference dimension). Added with respect to the original paper to loose the dependence of beta from alpha.