est.danco {Rdimtools}

Intrinsic Dimensionality Estimation with DANCo

Description

DANCo exploits the balanced information of both the normalized nearest neighbor distances as well as the angles of data pairs in the neighboring points.

Usage

est.danco(X, k = 5)

Arguments

X

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

k

the neighborhood size used for estimating local intrinsic dimension.

Value

a named list containing containing

estdim

estimated dimension via the method.

References

Ceruti C, Bassis S, Rozza A, Lombardi G, Casiraghi E, Campadelli P (2014). “DANCo: An Intrinsic Dimensionality Estimator Exploiting Angle and Norm Concentration.” Pattern Recognition, 47(8), 2569–2581.

Examples

## create 3 datasets of intrinsic dimension 2.
X1 = aux.gensamples(n=50, dname="swiss")
X2 = aux.gensamples(n=50, dname="ribbon")
X3 = aux.gensamples(n=50, dname="saddle")

## acquire an estimate for intrinsic dimension
out1 = est.danco(X1, k=10)
out2 = est.danco(X2, k=10)
out3 = est.danco(X3, k=10)

## print the results
line1 = paste0("* est.danco : 'swiss'  estiamte is ",round(out1$estdim,2))
line2 = paste0("* est.danco : 'ribbon' estiamte is ",round(out2$estdim,2))
line3 = paste0("* est.danco : 'saddle' estiamte is ",round(out3$estdim,2))
cat(paste0(line1,"\n",line2,"\n",line3))