arc (version 1.4.2)

cba_manual: CBA Classifier from provided rules

Description

Learns a CBA rule set from supplied rules

Usage

cba_manual(
  datadf_raw,
  rules,
  txns,
  rhs,
  classAtt,
  cutp,
  pruning_options = list(input_list_sorted_by_length = FALSE)
)

Value

Object of class CBARuleModel.

Arguments

datadf_raw

a data frame with raw data (numeric attributes are not discretized).

rules

Rules class instance output by the apriori package

txns

Transactions class instance passed to the arules method invocation. Transactions are created over discretized data frame - numeric values are replaced with intervals such as "(13;45]".

rhs

character vectors giving the labels of the items which can appear in the RHS ($rhs element of the APappearance class instance passed to the arules call)

classAtt

the name of the class attribute.

cutp

list of cutpoints used to discretize data (required for application of the model on continuous data)

pruning_options

custom options for the pruning algorithm overriding the default values.

Examples

  data(humtemp)
  data_raw<-humtemp
  data_discr <- humtemp

  #custom discretization
  data_discr[,1]<-cut(humtemp[,1],breaks=seq(from=15,to=45,by=5))
  data_discr[,2]<-cut(humtemp[,2],breaks=c(0,40,60,80,100))

  #change interval syntax from (15,20] to (15;20], which is required by MARC
  data_discr[,1]<-as.factor(unlist(lapply(data_discr[,1], function(x) {gsub(",", ";", x)})))
  data_discr[,2]<-as.factor(unlist(lapply(data_discr[,2], function(x) {gsub(",", ";", x)})))
  data_discr[,3] <- as.factor(humtemp[,3])

  #mine rules
  classAtt="Class"
  appearance <- getAppearance(data_discr, classAtt)
  txns_discr <- as(data_discr, "transactions")
  rules <- apriori(txns_discr, parameter =
   list(confidence = 0.5, support= 3/nrow(data_discr), minlen=1, maxlen=5), appearance=appearance)
  inspect(rules)


  rmCBA <- cba_manual(data_raw,  rules, txns_discr, appearance$rhs,
  classAtt, cutp= list(), pruning_options=NULL)
  inspect (rmCBA@rules)
  prediction <- predict(rmCBA,data_discr,discretize=FALSE)
  acc <- CBARuleModelAccuracy(prediction, data_discr[[classAtt]])
  print(paste("Accuracy:",acc))