RWeka: Interfaces for RWeka package for data science pipelines.

In rbertolusso/intubate: Interface to Popular R Functions for Data Science Pipelines

Description

Interfaces to RWeka functions that can be used in a pipeline implemented by magrittr.

Usage

ntbt_AdaBoostM1(data, ...)
ntbt_Bagging(data, ...)
ntbt_CostSensitiveClassifier(data, ...)
ntbt_DecisionStump(data, ...)
ntbt_Discretize(data, ...)
ntbt_GainRatioAttributeEval(data, ...)
ntbt_IBk(data, ...)
ntbt_InfoGainAttributeEval(data, ...)
ntbt_J48(data, ...)
ntbt_JRip(data, ...)
ntbt_LBR(data, ...)
ntbt_LogitBoost(data, ...)
ntbt_LinearRegression(data, ...)
ntbt_LMT(data, ...)
ntbt_Logistic(data, ...)
ntbt_M5P(data, ...)
ntbt_M5Rules(data, ...)
ntbt_MultiBoostAB(data, ...)
ntbt_Normalize(data, ...)
ntbt_OneR(data, ...)
ntbt_PART(data, ...)
ntbt_SMO(data, ...)
ntbt_Stacking(data, ...)

Arguments

data	data frame, tibble, list, ...
...	Other arguments passed to the corresponding interfaced function.

Details

Interfaces call their corresponding interfaced function.

Value

Object returned by interfaced function.

Author(s)

Roberto Bertolusso

Examples

## Not run: 
library(intubate)
library(magrittr)
library(RWeka)

## R/Weka Attribute Evaluators
## Original function to interface
GainRatioAttributeEval(Species ~ . , data = iris)
InfoGainAttributeEval(Species ~ . , data = iris)

## The interface puts data as first parameter
ntbt_GainRatioAttributeEval(iris, Species ~ .)
ntbt_InfoGainAttributeEval(iris, Species ~ .)

## so it can be used easily in a pipeline.
iris %>%
  ntbt_GainRatioAttributeEval(Species ~ .)
iris %>%
  ntbt_InfoGainAttributeEval(Species ~ .)

## R/Weka Classifier Functions
data(infert)
infert$STATUS <- factor(infert$case, labels = c("control", "case"))

## Original function to interface
LinearRegression(weight ~ feed, data = chickwts)
Logistic(STATUS ~ spontaneous + induced, data = infert)
SMO(Species ~ ., data = iris, control = Weka_control(K = list("RBFKernel", G = 2)))

## The interface puts data as first parameter
ntbt_LinearRegression(chickwts, weight ~ feed)
ntbt_Logistic(infert, STATUS ~ spontaneous + induced)
ntbt_SMO(iris, Species ~ ., control = Weka_control(K = list("RBFKernel", G = 2)))

## so it can be used easily in a pipeline.
chickwts %>%
  ntbt_LinearRegression(weight ~ feed)
infert %>%
  ntbt_Logistic(STATUS ~ spontaneous + induced)
iris %>%
  ntbt_SMO(Species ~ ., control = Weka_control(K = list("RBFKernel", G = 2)))

## R/Weka Lazy Learners
## No examples provided. LBR seems to need 'lazyBayesianRules'
## and I am too lazy myself to install it
ntbt_IBk(chickwts, weight ~ feed)   ## Example may not make sense


## R/Weka Meta Learners
## MultiBoostAB needs Weka package 'multiBoostAB'
## CostSensitiveClassifier throws an error

## Original function to interface
AdaBoostM1(Species ~ ., data = iris, control = Weka_control(W = "DecisionStump"))
Bagging(Species ~ ., data = iris, control = Weka_control())
LogitBoost(Species ~ ., data = iris, control = Weka_control())
Stacking(Species ~ ., data = iris, control = Weka_control())

## The interface puts data as first parameter
ntbt_AdaBoostM1(iris, Species ~ ., control = Weka_control(W = "DecisionStump"))
ntbt_Bagging(iris, Species ~ ., control = Weka_control())
ntbt_LogitBoost(iris, Species ~ ., control = Weka_control())
ntbt_Stacking(iris, Species ~ ., control = Weka_control())

## so it can be used easily in a pipeline.
iris %>%
  ntbt_AdaBoostM1(Species ~ ., control = Weka_control(W = "DecisionStump"))
iris %>%
  ntbt_Bagging(Species ~ ., control = Weka_control())
iris %>%
  ntbt_LogitBoost(Species ~ ., control = Weka_control())
iris %>%
  ntbt_Stacking(Species ~ ., control = Weka_control())

## R/Weka Rule Learners
## Original function to interface
JRip(Species ~ ., data = iris)
M5Rules(mpg ~ ., data = mtcars)
OneR(Species ~ ., data = iris)
PART(Species ~ ., data = iris)

## The interface puts data as first parameter
ntbt_JRip(iris, Species ~ .)
ntbt_M5Rules(mtcars, mpg ~ .)
ntbt_OneR(iris, Species ~ .)
ntbt_PART(iris, Species ~ .)

## so it can be used easily in a pipeline.
iris %>%
  ntbt_JRip(Species ~ .)
mtcars %>%
  ntbt_M5Rules(mpg ~ .)
iris %>%
  ntbt_OneR(Species ~ .)
iris %>%
  ntbt_PART(Species ~ .)

## R/Weka Classifier Trees
DF3 <- read.arff(system.file("arff", "cpu.arff", package = "RWeka"))
DF4 <- read.arff(system.file("arff", "weather.arff", package = "RWeka"))

## Original function to interface
DecisionStump(play ~ ., data = DF4)
J48(Species ~ ., data = iris)
LMT(play ~ ., data = DF4)
M5P(class ~ ., data = DF3)

## The interface puts data as first parameter
ntbt_DecisionStump(DF4, play ~ .)
ntbt_J48(iris, Species ~ .)
ntbt_LMT(DF4, play ~ .)
ntbt_M5P(DF3, class ~ .)

## so it can be used easily in a pipeline.
DF4 %>%
  ntbt_DecisionStump(play ~ .)
iris %>%
  ntbt_J48(Species ~ .)
DF4 %>%
  ntbt_LMT(play ~ .)
DF3 %>%
  ntbt_M5P(class ~ .)

## R/Weka Filters
w <- read.arff(system.file("arff","weather.arff", package = "RWeka"))

## Original function to interface
Discretize(play ~., data = w)
Normalize(~., data = w)

## The interface puts data as first parameter
ntbt_Discretize(w, play ~.)
ntbt_Normalize(w, ~.)

## so it can be used easily in a pipeline.
w %>%
  ntbt_Discretize(play ~.)
w %>%
  ntbt_Normalize(~.)

## End(Not run)

rbertolusso/intubate documentation built on May 27, 2019, 3 a.m.