Bug in Weight by Chi Squared Statistic

yzan

Whenever an exampleSet with a date data type is passed to "Weight by Chi Squared Statistic" operator, I get java.lang.ArrayIndexOutOfBoundsException. Since the operator can automatically deal with numeric attributes (by discretizing them) and similar operators like Information Gain Ratio or Gini can deal correctly date attributes, I am inclined to believe that this is a bug. The operator should be tested on other data types (like date_time and time) as well.

 

Provisional workaround: Convert the date to numeric with "Date to Numerical" operator.

 

Stack trace:

Exception: java.lang.ArrayIndexOutOfBoundsException
Message: 2147483647
Stack trace:

  com.rapidminer.operator.features.weighting.ChiSquaredWeighting.calculateWeights(ChiSquaredWeighting.java:119)
  com.rapidminer.operator.features.weighting.AbstractWeighting.doWork(AbstractWeighting.java:149)
  com.rapidminer.operator.Operator.execute(Operator.java:1004)
  com.rapidminer.operator.execution.SimpleUnitExecutor.execute(SimpleUnitExecutor.java:77)
  com.rapidminer.operator.ExecutionUnit$3.run(ExecutionUnit.java:812)
  com.rapidminer.operator.ExecutionUnit$3.run(ExecutionUnit.java:807)
  java.security.AccessController.doPrivileged(Native Method)
  com.rapidminer.operator.ExecutionUnit.execute(ExecutionUnit.java:807)
  com.rapidminer.operator.OperatorChain.doWork(OperatorChain.java:428)
  com.rapidminer.operator.Operator.execute(Operator.java:1004)
  com.rapidminer.Process.run(Process.java:1315)
  com.rapidminer.Process.run(Process.java:1179)
  com.rapidminer.Process.run(Process.java:1132)
  com.rapidminer.Process.run(Process.java:1127)
  com.rapidminer.Process.run(Process.java:1117)
  com.rapidminer.gui.ProcessThread.run(ProcessThread.java:65)

IngoRM

Thanks for pointing it out!  We will have a look into this.

Cheers,

Ingo