The class `"modelMatrix"`

and notably its subclass
`"dsparseModelMatrix"`

are used to encode additional information,
analogously to what the standard **R** function
`model.matrix()`

returns.

Only `"dsparseModelMatrix"`

and `"ddenseModelMatrix"`

are
“actual” (aka non-virtual) classes. For these,
objects can be created by calls of the form
`new("dsparseModelMatrix", x, assign, contrast)`

, where `x`

is a `dgCMatrix`

classed object.

The `"modelMatrix"`

mother class contains
`Matrix`

plus two extra slots,

`assign`

:`"integer"`

vector of length`ncol(.)`

, coding the variables which make up the matrix columns, see`model.matrix`

.`contrasts`

:a named

`list`

of`contrasts`

, as in`model.matrix()`

.`Dim`

:integer vector of length two with the matrix dimensions.

`Dimnames`

:list of length two, the

`dimnames(.)`

of the matrix.

whereas the (current only) actual classes `"d*ModelMatrix"`

,
have an at least an additional (`numeric`

slot `"x"`

.
E.g., "dsparseModelMatrix" has the additional slots

`i`

,`p`

:row number and “pointer” integer vectors, see class

`"dgCMatrix"`

.`x`

:`"numeric"`

vector of non-zero entries.`factors`

:a (possibly empty)

`list`

of factorizations.

`"dsparseModelMatrix"`

extends class `"dgCMatrix"`

directly,

`"ddenseModelMatrix"`

extends class `"dgeMatrix"`

directly.

- show
`signature(object = "modelMatrix")`

:`show(.)`

the matrix, but also the`assign`

and`contrasts`

slots.`signature(x = "modelMatrix")`

: as`show()`

, however (via`...`

) allowing to pass further arguments for printing the matrix.

Martin Maechler

`sparse.model.matrix`

will return a
`"dsparseModelMatrix"`

object.
`model.Matrix`

which is a simple wrapper around the traditional
`model.matrix`

and returns a `"ddenseModelMatrix"`

object.

1 2 3 4 |

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.

All documentation is copyright its authors; we didn't write any of that.