daCoDa | R Documentation |

Linear and quadratic discriminant analysis for compositional data using either robust or classical estimation.

```
daCoDa(x, grp, coda = TRUE, method = "classical", rule = "linear", ...)
```

`x` |
a matrix or data frame containing the explanatory variables |

`grp` |
grouping variable: a factor specifying the class for each observation. |

`coda` |
TRUE, when the underlying data are compositions. |

`method` |
“classical” or “robust” |

`rule` |
a character, either “linear” (the default) or “quadratic”. |

`...` |
additional arguments for the functions passed through |

Compositional data are expressed in orthonormal (ilr) coordinates (if `coda==TRUE`

). For linear
discriminant analysis the functions LdaClassic (classical) and Linda (robust) from the
package rrcov are used. Similarly, quadratic discriminant analysis
uses the functions QdaClassic and QdaCov (robust) from the same package.

The classical linear and quadratic discriminant rules are invariant to ilr coordinates and clr coefficients. The robust rules are invariant to ilr transformations if affine equivariant robust estimators of location and covariance are taken.

An S4 object of class LdaClassic, Linda, QdaClassic or QdaCov. See package rrcov for details.

Jutta Gamper

Filzmoser, P., Hron, K., Templ, M. (2012)
Discriminant analysis for compositional data and robust parameter estimation.
*Computational Statistics*, 27(4), 585-604.

`LdaClassic`

, `Linda`

,
`QdaClassic`

, `QdaCov`

```
## toy data (non-compositional)
require(MASS)
x1 <- mvrnorm(20,c(0,0,0),diag(3))
x2 <- mvrnorm(30,c(3,0,0),diag(3))
x3 <- mvrnorm(40,c(0,3,0),diag(3))
X <- rbind(x1,x2,x3)
grp=c(rep(1,20),rep(2,30),rep(3,40))
clas1 <- daCoDa(X, grp, coda=FALSE, method = "classical", rule="linear")
summary(clas1)
## predict runs only with newest verison of rrcov
## Not run:
predict(clas1)
## End(Not run)
# specify different prior probabilities
clas2 <- daCoDa(X, grp, coda=FALSE, prior=c(1/3, 1/3, 1/3))
summary(clas2)
## compositional data
data(coffee)
x <- coffee[coffee$sort!="robusta",2:7]
group <- droplevels(coffee$sort[coffee$sort!="robusta"])
cof.cla <- daCoDa(x, group, method="classical", rule="quadratic")
cof.rob <- daCoDa(x, group, method="robust", rule="quadratic")
## predict runs only with newest verison of rrcov
## Not run:
predict(cof.cla)@ct
predict(cof.rob)@ct
## End(Not run)
```

Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.