Guidance for weka CDT

(Saisai Ma, 06/09/2015)

CDT: a Program for generating Causal Decision Trees. CDT algorithm is located at weka-classifiers-causality

Parameter settings

• height: The maximum height of the causal decision tree. (default: 5)
• pruned: Whether pruning is performed. (default: True)
• subclass: Whether subclassification (propensity score) is performed, exact matching otherwise. (default: False)
  In this version, subclassification is performed by invoking the function (formula, data, method=“subclass”, …) in the R package MatchIt. Thus extra configurations need to be done:
  
  • Run the following commands in R (Note that keep Rserve() running when using “subclass” option)
    

    > install.packages("MatchIt")
    > install.packages("Rserve")
    > library(Rserve)
    > Rserve()

• location: The location of propensityTest_weka.R file. When subclassification selected, please indicate the location of propensityTest_weka.R. This R file is employed to do subclassfication and return the subclasses to Java. This option only works when subclass=True.
• numControl: The maximum number of control list during the causal testing. If the number of control list is larger than the threshold L, then select the first L variables as the control. This option only works when subclass=False.

Download

To download the program, please click the followong link: Causal Decision Tree implementation

Last Revised: April 2022