Help for package ModelMatrixModel

Type:

Package

Title:

Create Model Matrix and Save the Transforming Parameters

Version:

0.1.0

Author:

Xinyong Tian

Maintainer:

Xinyong Tian <xinyongtian@gmail.com>

Description:

The model.matrix() function in R is convenient for transforming training dataset for modeling. But it does not save any parameter used in transformation, so it is hard to apply the same transformation to test dataset or new dataset. This package is created to solve the problem.

License:

GPL-3

Encoding:

UTF-8

LazyData:

false

RoxygenNote:

7.1.1

Imports:

Matrix

Suggests:

rmarkdown, knitr

VignetteBuilder:

knitr

NeedsCompilation:

Packaged:

2021-08-16 16:52:10 UTC; C07520

Repository:

CRAN

Date/Publication:

2021-08-17 08:00:10 UTC

ModelMatrixModel() function

Description

This function transforms a data.frame to matrix (or sparse matrix) based on a r formula. The mean different from model.matrix() function is that it outputs a class stored with the transformed matrix, as well as the transforming parameters which can be applied to new data.

Usage

ModelMatrixModel(
  rformula,
  data,
  sparse = TRUE,
  center = FALSE,
  scale = FALSE,
  remove_1st_dummy = FALSE,
  verbose = FALSE
)

Arguments

rformula

a formula, e.g. formula("~ 1+x1+x2"),"~ 1+x1+x2",or ~ 1+x1+x2 . Note the interpreting of the formula might be different slightly from model.matrix function. In model.matrix(),intercept column will be included in output matrix with or without "1" in the formula. But in ModelMatrixModel(),intercept column will be included in output matrix only when "1" is present. Moreover "0" or "." in the formula will be ignored.

data

a data.frame.

sparse

boolean, if TRUE return a sparse matrix, i.e. a "dgCMatrix" class.

center

boolean, if center the output.

scale

boolean, if scale the output.

remove_1st_dummy

boolean, if remove the first dummy variable in one hot key transformation.

verbose

boolean, if print out progress.

Details

see vignettes.

Value

A ModelMatrixModel class,which includes the transformed matrix and the transforming parameters.

Examples

library(ModelMatrixModel)
traindf= data.frame(x1 = sample(LETTERS[1:5], replace = TRUE, 20),
                    x2 = rnorm(20, 100, 5),
                    y = rnorm(20, 10, 2))
mm=ModelMatrixModel(~x1+x2,traindf,remove_1st_dummy = FALSE)
data.frame(as.matrix(head(mm$x,2)))

predict() function

Description

This function transforms new data based on transforming parameters from a ModelMatrixModel object

Usage

## S3 method for class 'ModelMatrixModel'
predict(object, data, handleInvalid = "keep", verbose = FALSE, ...)

Arguments

object

a ModelMatrixModel object.

data

a data.frame.

handleInvalid

a string,'keep' or 'error'. In dummy variable transformation, if categorical variable has a factor level that is unseen before, 'keep' will keep the record, output dummy variables will be all zero.

verbose

boolean, if print out progress.

...

other parameters.

Value

A ModelMatrixModel class,which includes the transformed matrix and the necessary transforming parameters copied from input object.

Examples

library(ModelMatrixModel)
traindf= data.frame(x1 = sample(LETTERS[1:5], replace = TRUE, 20),
                    x2 = rnorm(20, 100, 5),
                    y = rnorm(20, 10, 2))
newdf=data.frame(x1 = sample(LETTERS[1:5], replace = TRUE, 3),
                 x2 = rnorm(3, 100, 5))
mm=ModelMatrixModel(~x1+x2,traindf,remove_1st_dummy = FALSE)
mm_pred=predict(mm,newdf)
data.frame(as.matrix(head(mm_pred$x,2)))