do.sdlpp: Sample-Dependent Locality Preserving Projection

Description

Many variants of Locality Preserving Projection are contingent on graph construction schemes in that they sometimes return a range of heterogeneous results when parameters are controlled to cover a wide range of values. This algorithm takes an approach called sample-dependent construction of graph connectivity in that it tries to discover intrinsic structures of data solely based on data.

Usage

do.sdlpp(
  X,
  ndim = 2,
  t = 1,
  preprocess = c("center", "scale", "cscale", "decorrelate", "whiten")
)

Arguments

an \((n\times p)\) matrix or data frame whose rows are observations.

ndim

an integer-valued target dimension.

kernel bandwidth in \((0,\infty)\).

preprocess

an additional option for preprocessing the data. Default is "center". See also aux.preprocess for more details.

Value

a named list containing

Y: an \((n\times ndim)\) matrix whose rows are embedded observations.
trfinfo: a list containing information for out-of-sample prediction.
projection: a \((p\times ndim)\) whose columns are basis for projection.

References

yang_sampledependent_2010Rdimtools

Examples

Run this code

# NOT RUN {
## use iris data
data(iris)
X     = as.matrix(iris[,1:4])
label = as.integer(iris$Species)

## compare with PCA
out1 <- do.pca(X,ndim=2)
out2 <- do.sdlpp(X, t=0.1)
out3 <- do.sdlpp(X, t=1)
out4 <- do.sdlpp(X, t=10)

## visualize
opar <- par(no.readonly=TRUE)
par(mfrow=c(2,2))
plot(out1$Y, col=label, main="PCA")
plot(out2$Y, col=label, main="SDLPP::t=0.1")
plot(out3$Y, col=label, main="SDLPP::t=1")
plot(out4$Y, col=label, main="SDLPP::t=10")
par(opar)
# }
# NOT RUN {

# }