ks (version 1.6.13)

kda.kde: Kernel density estimate for kernel discriminant analysis for multivariate data

Description

Kernel density estimate for kernel discriminant analysis for 1- to 6-dimensional data

Usage

kda.kde(x, x.group, Hs, hs, prior.prob=NULL, gridsize, xmin, xmax,
        supp=3.7, eval.points=NULL, binned=FALSE, bgridsize, w,
        compute.cont=FALSE, approx.cont=TRUE)

Arguments

x

matrix of training data values

x.group

vector of group labels for training data

Hs

(stacked) matrix of bandwidth matrices

hs

vector of scalar bandwidths

prior.prob

vector of prior probabilities

gridsize

vector of number of grid points

xmin

vector of minimum values for grid

xmax

vector of maximum values for grid

supp

effective support for standard normal is [-supp, supp]

eval.points

points at which density estimate is evaluated

binned

flag for binned kernel estimation

bgridsize

vector of binning grid sizes - only required if binned=TRUE

w

vector of weights (non-negative and sum is equal to sample size)

compute.cont

flag for computing probability contour levels from 1% to 99%

approx.cont

flag for computing approximate probability contour levels

Value

• The kernel density estimate for kernel discriminant analysis is based on kde, one density estimate for each group.

  The result from kda.kde is a density estimate for discriminant analysis is an object of class kda.kde which is a list with 6 fields

• x
  data points - same as input
• x.group
  group labels - same as input
• eval.points
  points that density estimate is evaluated at
• estimate
  density estimate at eval.points
• prior.prob
  prior probabilities
• H
  bandwidth matrices (>1-d only) or
• h
  bandwidths (1-d only)
• cont
  probability contour levels
• w
  weights

Details

For d = 1, 2, 3, 4, and if eval.points is not specified, then the density estimate is computed over a grid defined by gridsize (if binned=FALSE) or by bgridsize (if binned=TRUE).

For d = 1, 2, 3, 4, and if eval.points is specified, then the density estimate is computed is computed exactly at eval.points. For d > 4, the kernel density estimate is computed exactly and eval.points must be specified.

If you have prior probabilities then set prior.prob to these. Otherwise prior.prob=NULL is the default i.e. use the sample proportions as estimates of the prior probabilities.

The default xmin is min(x) - Hmax*supp and xmax is max(x) + Hmax*supp where Hmax is the maximim of the diagonal elements of H.

The default weights w is a vector of all ones.

References

Wand, M.P. & Jones, M.C. (1995) Kernel Smoothing. Chapman & Hall. London.

See Also

plot.kda.kde

Examples

### See examples in ? plot.kda.kde