finished significant code overhaul

[David]

Hey Venki,
Great news, I am mostly finished with an overhaul of inqle. I removed
a layer of unneeded complexity (the LearningCycle), improving
usability. More importantly, I significantly simplified the process
to contribute new RapidMiner experiments.

So these instructions supersede those I gave you before.
To contribute your new rules-based RM experiment, you will need to:

(1) Make plugin.xml look like this

<?xml version="1.0" encoding="UTF-8"?>
<?eclipse version="3.2"?>
<plugin>
<extension

point="org.inqle.experiment.rapidminer.IRapidMinerExperiment">
<rapidminer-experiment

id="org.inqle.experiment.rapidminer.rule.VenkisRulesExperiment"
class="org.inqle.experiment.rapidminer.rule.
VenkisRulesExperiment"
file="/org/inqle/experiment/rapidminer/rules/xml/
VenkisRulesExperiment.xml"
name="Venkis Rule Learner"
description="whatever"
type="rules"
weight="10000">
</rapidminer-experiment>
</extension>
</plugin>

(2) Place your RM experiment on the classpath, and ensure that the
file="..."
above points to it.

(3) Create your new experimenter class. Must implement interface
IRapidMinerExperiment. Just start with class
ClassificationRegressionCrossValidationExperiment and modify it
Ensure that
class="..." above points to it.
Note that you will have to generate a new class of result. This class
must implement IExperimentResult. I suggest you might extend class
PerformanceVectorResult, since the output of your rules experiment
would include a performance vector containing its accuracy,
correlation, etc.?

You might be done there. The ExperimenterAgentWizard should allow the
user to select your experiment. Or your experiment will be selected
at random if the ExperimenterAgentWizard is not customized to use a
particular experiment.

I need to do some more work. I am sure there are bugs. I'd like to
make some of the wizard pages a little nicer.

Dave

[David]

A update

Disregard this "I suggest you might extend class

PerformanceVectorResult, since the output of your rules experiment
would include a performance vector containing its accuracy,

correlation, etc.? ". You will make a totally new Jenabean class
implementing IExperimentResult.