denpro (version 0.9.2)

sim.data: Generates data for illustrative purposes

Description

Returns a random sample from some distributions, to illustrate some visulization tools. Returns also the density (as a piecewise constant function) for some examples, or the distribution function.

Usage

sim.data(n = NULL, seed = 1, N = NULL, type = "mulmod",
M = NULL, sig = NULL, p = NULL, d = NULL,
cova = NULL, marginal = NULL, t = NULL, df = NULL, distr = FALSE,
noisedim = 1, sig1 = 0.5, sig2 = 1.5, diff = 0.1, dist = 4)

Arguments

n

positive integer; size of the sample to be generated

seed

real number; seed for the random number generator.

N

2*1 vector of positive integers; the size of the grid where the piecewise constant function is evaluated

type

"mixt", "mulmod", "fox", "tetra3d", "penta4d", "cross", "gauss", "student", "gumbel", "1d2modal", or "claw".

M

mixnum*d-matrix; rows of M are means of the Gaussians in the mixture. We have a mixture of "mixnum" Gaussians, whose dimension is d.

sig

mixnum*d-matrix; rows of sig are the diagonals of the covariance matrices of the mixtures.

p

mixnum-vector; weights for the members of the mixture. The sum of elements of "p" is 1.

d

positive integer; dimension of the vectors of the sample to be generated, need to be given only when type="mixt" and d=1

cova

Covariance matrix for the Gauss or Student copulas

marginal

NULL, "gauss", or "student"; this parameter is used to give the marginal distribution for the Gauss or Student copulas; if marginal=NULL, then the uniform marginals are used

t

if marginal="student", gives the degrees of freedom

df

degrees of freedom for the Student copula

distr

internal (implemented for "1d2modal") TRUE, if one wants the distribution function instead of the density function

noisedim

the number of noise dimension in the projection pursuit example ("fssk")

sig1

standard deviation for "cross" and "diff1d"

sig2

second standard deviation for "cross"

diff

parameter for "diff1d"; the difference between the Gaussians in the 1D mixture

dist

a positive real number; gives the distance between the mixture centers in the 4D mixture of Gaussians "penta4d"

Value

If "n" is not NULL, then the function returns a n*d-data matrix or a n*2-data matrix, if "N" is not NULL, then the function returns a piecewise constant function on the grid of size N[1]*N[2], if the both are NULL, then the function returns the mean, covariance, and the weights of the mixture components

Details

When type="mixt", generates data from a mixture of Gaussians. When type="mulmod", the density is 3-modal. When type="fox", the density has multimodal level sets.

Examples

d<-2
mixnum<-3
M<-matrix(0,mixnum,d)
M[1,]<-c(0,0)   
M[2,]<-c(4,0)  
M[3,]<-c(0,4)   
sig<-matrix(1,mixnum,d)
p0<-1/mixnum
p<-p0*rep(1,mixnum)
n<-100
dendat<-sim.data(type="mixt",n=n,M=M,sig=sig,p=p,seed=1)    
plot(dendat)

dendat<-sim.data(n=100)
plot(dendat)

N<-c(20,20)
pcf<-sim.data(N=N)
dp<-draw.pcf(pcf,pnum=c(30,30))
contour(dp$x,dp$y,dp$z,drawlabels=FALSE)

sim.data()

type="fox"
dendat<-sim.data(n=100,type=type)
plot(dendat)

pcf<-sim.data(N=N,type=type)
dp<-draw.pcf(pcf,pnum=c(30,30))
contour(dp$x,dp$y,dp$z,drawlabels=FALSE)