compositions (version 2.0-4)

groupparts: Group amounts of parts

Description

Groups parts by amalgamation or balancing of their amounts or proportions.

Usage

groupparts(x,...)
# S3 method for acomp
groupparts(x,...,groups=list(...))
# S3 method for rcomp
groupparts(x,...,groups=list(...))
# S3 method for aplus
groupparts(x,...,groups=list(...))
# S3 method for rplus
groupparts(x,...,groups=list(...))
# S3 method for ccomp
groupparts(x,...,groups=list(...))

Value

a new dataset of the same type with each group represented by a single column

Arguments

x

an amount/compositional dataset

...

further parameters to use (actually ignored)

groups

a list of numeric xor character vectors, each giving a group of parts

Missing Policy

For the real geometries, SZ and BDL are considered as 0, and MAR and MNAR are kept as missing of the same type. For the relative geometries, a BDL is a special kind of MNAR, whereas a SZ is qualitatively different (thus a balance with a SZ has no sense). MAR values transfer their MAR property to the resulting new variable.

Author

K.Gerald v.d. Boogaart http://www.stat.boogaart.de, Raimon Tolosana-Delgado

Details

In the real geometry grouping is done by amalgamation (i.e. adding the parts). In the Aitchison-geometry grouping is done by taking geometric means. The new parts are named by named formal arguments. Not-mentioned parts remain ungrouped.

References

Egozcue, J.J. and V. Pawlowsky-Glahn (2005) Groups of Parts and their Balances in Compositional Data Analysis, Mathematical Geology, in press

See Also

aplus

Examples

data(SimulatedAmounts)
plot(groupparts(acomp(sa.lognormals5),A=c(1,2),B=c(3,4),C=5))
plot(groupparts(aplus(sa.lognormals5),B=c(3,4),C=5))
plot(groupparts(rcomp(sa.lognormals5),A=c("Cu","Pb"),B=c(2,5)))
hist(groupparts(rplus(sa.lognormals5),1:5))