| eval.stage.gauss {delt} | R Documentation |
Estimates a 1D density with a mixture of Gaussians. The mixture is found by minimizing the L2 empirical risk in a stagewise manner.
eval.stage.gauss(dendat, M, mugrid, siggrid = 1, sigeka = TRUE, src = "c", sampstart=FALSE, boost=FALSE, N=60)
dendat |
n-vector of 1D observations |
M |
integer >= 1; the number of mixture components in the estimate |
mugrid |
a vector of real numbers; the range for the means of mixture components |
siggrid |
a vector of real numbers; the range of possible standard deviations in the mixture components |
sigeka |
TRUE or FALSE; if TRUE, then the standard deviation of the first mixture component is equal to 1, otherwise the standard deviation of the first mixture component is found by minimization |
src |
"R" or "c"; if "R", then the R-code is used, otherwise the c-code is used |
sampstart |
internal |
boost |
internal |
N |
postive integer; the number of evaluation points |
A piecewise constant function with the additional components:
muut |
vector of real numbers; the means of the mixture components |
sigit |
vector of positive real numbers; the standard deviations of the mixture components |
curmix |
a probability vector; the weights of the mixture components |
Jussi Klemela
Jussi Klemela (2005). Density Estimation with Stagewise Optimization of the Empirical Risk
dendat<-sim.data(n=100,type="1d2modal",seed=1)
mugrid<-seq(-1,5,0.3) # grid of mu-values
siggrid<-seq(0.2,2,0.2) # grid of sigma-values
M<-17 # number of mixture components
pcf<-eval.stage.gauss(dendat,M,mugrid,siggrid)
dp<-draw.pcf(pcf)
plot(dp$x,dp$y,type="l")
# draw the estimate with the help of package "denpro"
#N<-100
#pcf2<-pcf.func("mixt",N,sig=pcf$sigit,M=pcf$muut,p=pcf$curmix)
#pnum<-100
#dm<-draw.pcf(pcf2,pnum=pnum)
#plot(dm$x,dm$y,type="l")