DPCalc: Calculate change-points using DP approach given number of...

In boris109able/ChangePointCalc: Estimate change-points via Sparse Group Lasso and Dynamic Programming.

Description

This function is a part of DP approach. This function calculates change-points using DP approach given number of change-points.

Usage

DPCalc(x, y, n, p, Kstar, c)

Arguments

x	input data x
y	input data y
n	total number of data
p	dimension of data
Kstar	the given number of change-points
c	cost matrix c (can be computed by ComputeCost() function)

Author(s)

Bingwen Zhang

References

Bingwen Zhang, Jun Geng and Lifeng Lai, Detecting changes in regression models via sparse group lasso, http://users.wpi.edu/~bzhang/icassp_v1.pdf

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (x, y, n, p, Kstar, c) 
{
    r <- matrix(0, ncol = n, nrow = Kstar + 1)
    s <- matrix(0, ncol = n, nrow = Kstar + 1)
    for (t in 1:n) {
        r[1, t] <- c[1, t]
        s[1, t] <- 1
    }
    for (K in 2:(Kstar + 1)) {
        for (t in K:n) {
            q = -1
            for (j in (K - 1):(t - 1)) {
                tmp <- r[K - 1, j] + c[j + 1, t]
                if (q > tmp || q < 0) {
                  q <- tmp
                  s[K, t] <- j + 1
                }
            }
            r[K, t] <- q
        }
    }
    return(list(r = r, s = s))
  }

boris109able/ChangePointCalc documentation built on May 13, 2019, 12:34 a.m.