discretize: Discretization

In FSelectorRcpp: 'Rcpp' Implementation of 'FSelector' Entropy-Based Feature Selection Algorithms with a Sparse Matrix Support

Discretization

Description

Discretize a range of numeric attributes in the dataset into nominal attributes. Minimum Description Length (MDL) method is set as the default control. There is also available equalsizeControl method.

Usage

discretize(
  x,
  y,
  control = list(mdlControl(), equalsizeControl()),
  all = TRUE,
  discIntegers = TRUE,
  call = NULL
)

mdlControl()

equalsizeControl(k = 10)

customBreaksControl(breaks)

Arguments

x	Explanatory continuous variables to be discretized or a formula.
y	Dependent variable for supervised discretization or a data.frame when x ia a formula.
control	discretizationControl object containing the parameters for discretization algorithm. Possible inputs are mdlControl or equalsizeControl, so far. If passed as a list, the first element is used.
all	Logical indicating if a returned data.frame should contain other features that were not discretized. (Example: should Sepal.Width be returned, when you pass iris and discretize Sepal.Length, Petal.Length, Petal.Width.)
discIntegers	logical value. If true (default), then integers are treated as numeric vectors and they are discretized. If false integers are treated as factors and they are left as is.
call	Keep as NULL. Inner method parameter for consistency.
k	Number of partitions.
breaks	custom breaks used for partitioning.

Author(s)

Zygmunt Zawadzki [e-mail]

References

U. M. Fayyad and K. B. Irani. Multi-Interval Discretization of Continuous-Valued Attributes for Classification Learning. In 13th International Joint Conference on Uncertainly in Artificial Intelligence(IJCAI93), pages 1022-1029, 1993.

Examples

# vectors
discretize(x = iris[[1]], y = iris[[5]])

# list and vector
head(discretize(x = list(iris[[1]], iris$Sepal.Width), y = iris$Species))

# formula input
head(discretize(x = Species ~ ., y = iris))
head(discretize(Species ~ ., iris))

# use different methods for specific columns
ir1 <- discretize(Species ~ Sepal.Length, iris)
ir2 <- discretize(Species ~ Sepal.Width, ir1, control = equalsizeControl(3))
ir3 <- discretize(Species ~ Petal.Length, ir2, control = equalsizeControl(5))
head(ir3)

# custom breaks
ir <- discretize(Species ~ Sepal.Length, iris,
  control = customBreaksControl(breaks = c(0, 2, 5, 7.5, 10)))
head(ir)

## Not run: 
# Same results
library(RWeka)
Rweka_disc_out <- RWeka::Discretize(Species ~ Sepal.Length, iris)[, 1]
FSelectorRcpp_disc_out <- FSelectorRcpp::discretize(Species ~ Sepal.Length,
                                                    iris)[, 1]
table(Rweka_disc_out, FSelectorRcpp_disc_out)
# But faster method
library(microbenchmark)
microbenchmark(FSelectorRcpp::discretize(Species ~ Sepal.Length, iris),
               RWeka::Discretize(Species ~ Sepal.Length, iris))


## End(Not run)

FSelectorRcpp documentation built on April 28, 2023, 5:07 p.m.