SGLPlotNorm: Plot change-points using norm by SGL approach

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

Description

Plot change-points using norm by SGL approach

Usage

SGLPlotL1Norm(fit, n, p, num = 1, breaks = NULL, norm = 2)

Arguments

fit	result of SGL approach
n	total number of data
p	dimension of data
num	the num'th lambda used by user
breaks	list of breaks where to put a tick on the x-axis
norm	norm=2 is l2 norm, norm=1 is l1 norm

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(fit, n, p, num=1, breaks=NULL,norm=2)
{
  # this function plot the l1 norm of theta in the result
  # num is the num th lambda we used
  # n is the number of data sample
  # p is the number of dimension
  #norm=2 is l2 norm, norm=1 is l1 norm
  #breaks is the break tick the user want to show on the plot, 
  #if breaks is NULL the default setting is to divide x-axis into 10 equal segment and label the 
  #tick accordingly
  t_sum = matrix(NA, n, 1)
  if(norm==1)
  {
    for (i in 1:n)
    { t_sum[i] = sum(abs(fit$beta[((i-1)*p+1):(i*p), num]))} 
  }
  if(norm==2)
  {
    for (i in 1:n)
    { t_sum[i] = sqrt(sum((fit$beta[((i-1)*p+1):(i*p), num])^2))}
  }
{
    if(is.null(breaks))
    {
      plotTsum + geom_point() + theme_bw(base_size=16) + xlab("Number of Observations") + ylab(expression(paste(group("|",group("|", theta[i], "|"),"|"))[1]))+scale_x_continuous(breaks=seq(1, n, floor(n/10)))
    }
    else
    {
      plotTsum + geom_point() + theme_bw(base_size=16) + xlab("Number of Observations") + ylab(expression(paste(group("|",group("|", theta[i], "|"),"|"))[1]))+scale_x_continuous(breaks=breaks)
    }
  }
  #plotTsum + geom_point() + theme_bw(base_size=16) + xlab("Number of Observations") + ylab(expression(paste(group("|",group("|", theta[i], "|"),"|"))[1]))+scale_x_continuous(breaks=seq(1, n, floor(n/10)))
}

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