xyz_search: Interaction search

In xyz: The 'xyz' Algorithm for Fast Interaction Search in High-Dimensional Data

Description

Interaction search

Usage

xyz_search(X, Y, L = 10, N = 100, binary = TRUE, negative = TRUE)

Arguments

X	A matrix.
Y	A vector.
L	An integer indicating how many projection steps are performed.
N	A integer, controlling the number of pairs that will be returned in the end.
binary	A logical indicating if X is binary or continuous.
negative	A logical indicating if also negative interactions should be searched for.

Value

N strongest interactions between X and Y after L projections.

References

G. Thanei, N. Meinshausen and R. Shah (2016). The xyz algorithm for fast interaction search in high-dimensional data. <https://arxiv.org/pdf/1610.05108v1.pdf>

Examples

n<-300
p<-1000
#construct a binary matrix
X<-matrix(sample(c(-1,1),replace=TRUE,n*p),n,p)
#set an interaction of the pair (1,2)
Y<-X[,1]*X[,2]+rnorm(n)
#run the interaction search
result<-xyz_search(X,Y,L=10,N=10,binary=TRUE,negative=TRUE)
#print the result
print(result)

Example output

intereaction pair: (1,2) strength: 1.048204
intereaction pair: (131,921) strength: 0.3533604
intereaction pair: (184,266) strength: 0.3187157
intereaction pair: (247,799) strength: 0.3093772
intereaction pair: (116,848) strength: 0.3082336
intereaction pair: (36,582) strength: 0.3061791
intereaction pair: (72,578) strength: 0.3061472
intereaction pair: (764,966) strength: 0.3056794
intereaction pair: (398,703) strength: 0.3006051
intereaction pair: (302,494) strength: 0.2994251

xyz documentation built on May 2, 2019, 10:25 a.m.