pollen {nnclust}R Documentation

Artificial "pollen-grain" data from 1986 Data Expo.

Description

Artificial data described as measurements of pollen grains. Created for the Data Expo at the 1986 Joint Statistical Meetings by David Coleman of RCA Labs.

Usage

data(pollen)

Format

A data frame with 3848 observations on the following 6 variables.

ridge
a numeric vector
nub
a numeric vector
crack
a numeric vector
weight
a numeric vector
density
a numeric vector
id
a numeric vector

Source

http://lib.stat.cmu.edu/data-expo/

Examples

data(pollen)
scaled <- scale(as.matrix(pollen[,1:5]))

## clearly at least two populations, based on neighbour distances
d <- nnfind(scaled)
plot(density(d$dist))
abline(v=0.08)

## cluster to extract two populations
nn <- nncluster(scaled, threshold=c(0.08,1), give.up=0)
nn

## tight cluster is the hidden message
plot(ridge~nub, data=pollen, subset=clusterMember(nn)==1)

## data set is a hollow ellipsoid with a treat in the middle.
coplot(density~ridge|crack*nub, data=pollen,pch=".",
    col=clusterMember(nn),cex=2, n=5)


[Package nnclust version 2.2 Index]