awojna / rseslib Goto Github PK

Hello,

I have a simple question, and I have probably overlooked something obvious: is there a way to extract the rules generated by a trained model?

weka.classifiers.rules.JRip has the useful method getRuleset(). Is there an equivalent for weka.classifiers.rules.RoughSet?

I know the rules can be visualised with Qmak, but I would like to extract the list of rules into a text based format for further analysis and optimisation.

Thanks,
Ben

Hello!

Currently, I am trying to generate rule-based models using the AQ15 algorithm. For some datasets, I encounter a NullPointerException. After some debugging, I have identified the issue.

I am using the rseslib library by importing the Java package. I create my model using the following function:

DoubleDataTable trainTable = new ArrayListDoubleDataTable(new File(trainPath), new EmptyProgress());
...
AQ15Classifier aq15 = new AQ15Classifier((Properties) null, trainTable, prog);

When running AQ15Classifier() I got the error. This creator runs generate(DoubleDataTable tab, Progress prog) from class CoveringRuleGenerator. In this generate function, the generate function from AQ15OneRuleGenerator is started.
In the function generate from AQ15OneRuleGenerator there is a code:

Rule rule = null;
if (candidates.size() > 0) {
   rule = new BooleanFunctionRule((BooleanFunction)candidates.get(0), e_pos.getDecision(), examples.attributes().nominalDecisionAttribute());
}

return rule;

For some dataset the candidates.size() is equal 0 and this function returns null. This value goes back to generate function from CoveringRuleGenerator, where there is a code:

for(int prooningObjts = uncovered.size(); (double)uncovered.size() > this.m_prooning * (double)prooningObjts; uncovered = this.removeCovered(uncovered, rules)) {
  rules.add(this.m_oneRuleGenerator.generate(tab, uncovered, this.m_k, dec));
}

For example, after some iterations I got rules object that looks like this:

We can see that the last added rule is null. Then, the function uncovered = this.removeCovered(uncovered, rules) is started (this execution is inside the above for loop). The removeCovered function is called with the rules object, which has a null value, resulting in a NullPointerException.

Can I do something that will eliminate the NullPointerException error?

I add simple code and dataset for which the problem occurs.

AQ15ErrorExample.zip