gSeg (version 0.1)

gSeg: Graph-Based Change-Point Detection

Description

This package can be used to estimate change-points in a sequence of observations, where the observation can be a vector or a data object, e.g., a network. A similarity graph is required. It can be a minimum spanning tree, a minimum distance pairing, a nearest neighbor graph, or a graph based on domain knowledge.

If you believe the sequence has at most one change point, the function gseg1 should be used. If you believe an interval of the sequence has a changed distribution, the function gseg2 should be used. If you feel the sequence has multiple change-points, you can use gseg1 and gseg2 multiple times. See gseg1 and gseg2 for the details of these two function.

References

Chen, H. and Zhang, N.R. (2014) Graph-Based Change-Point Detection.

Online access: http://arxiv.org/abs/1209.1625

See Also

gseg1, gseg2

Examples

data(Example) # this example data has: n (the number of observations) and E (an edge matrix)
r1 = gseg1(n,E)
r2 = gseg2(n,E)