Rapidminer changes my values...

Jorge

Hi,

I'm working with Rapidminer 4.3 in that project

<operator name="Root" class="Process" expanded="yes">
    <operator name="ArffExampleSource" class="ArffExampleSource" breakpoints="after">
        <parameter key="data_file"	value="C:\Input.arff"/>
        <parameter key="id_attribute"	value="1"/>
        <parameter key="label_attribute"	value="example6"/>
    </operator>
    <operator name="InteractiveAttributeWeighting" class="InteractiveAttributeWeighting">
    </operator>
    <operator name="Learn" class="OperatorChain" expanded="yes">
        <operator name="W-NaiveBayesUpdateable" class="W-NaiveBayesUpdateable">
        </operator>
        <operator name="ModelWriter" class="ModelWriter">
            <parameter key="model_file"	value="C:\model.mod"/>
            <parameter key="output_type"	value="XML"/>
        </operator>
    </operator>
    <operator name="ArffExampleSource (2)" class="ArffExampleSource" breakpoints="after">
        <parameter key="data_file"	value="C:\Prediction.arff"/>
        <parameter key="id_attribute"	value="1"/>
        <parameter key="label_attribute"	value="example6"/>
    </operator>
    <operator name="ModelLoader" class="ModelLoader">
        <parameter key="model_file"	value="C:\model.mod"/>
    </operator>
    <operator name="ModelApplier" class="ModelApplier">
    </operator>
</operator>

with input.arff...

@RELATION Input

@ATTRIBUTE Id numeric
@ATTRIBUTE example1 string
@ATTRIBUTE example2 string
@ATTRIBUTE example3 string
@ATTRIBUTE example4 string
@ATTRIBUTE example5 string
@ATTRIBUTE example6 string

@DATA
'1','ex1','hello4','hw1','false','1000k','slow'
'2','ex1','hello6','hw2','true','4000k','slow'
'3','ex1','hello2','hw3','false','500k','slow'
'4','ex1','hello3','hw3','true','2000k','slow'
'5','ex2','hello2','hw2','true','500k','slow'
'6','ex2','hello5','hw1','true','1000k','mid'
'7','ex2','hello2','hw3','false','4000k','fast'
'8','ex3','hello','hw1','true','2000k','mid'
'9','ex3','hello','hw2','true','4000k','fast'
'10','ex3','hello','hw3','false','2000k','slow'
'11','ex3','hello','hw1','false','500k','mid'

and prediction.arff.....

@RELATION Prediction

@ATTRIBUTE Id numeric
@ATTRIBUTE example1 string
@ATTRIBUTE example2 string
@ATTRIBUTE example3 string
@ATTRIBUTE example4 string
@ATTRIBUTE example5 string


@DATA
'100','ex1','hello','hw1','false','1000k'
'101','ex1','hello2','hw2','true','4000k'
'102','ex1','hello','hw2','true','4000k'
'103','ex1','hello2','hw3','true','500k'
'104','ex1','hello','hw2','true','2000k'
'105','ex1','hello2','hw1','true','4000k'
'106','ex2','hello3','hw1','false','500k'
'107','ex3','hello3','hw2','true','4000k'
'108','ex3','hello4','hw3','true','500k'
'109','ex3','hello5','hw3','false','500k'
'110','ex3','hello6','hw2','true','500k'
'111','ex3','hello2','hw1','false','500k'
'112','ex3','hello6','hw1','true','500k'

when I execute the program, at the results, I click on "Data View" of the "Data Table" and the values of the colum "example1" are differents of the prediction.arff example1 attribute.

Anyone can help me?
Is only a print error, or affects too in the learning operator?

Thanks in advance.

Cheers,
Jorge

steffen

Hello Jorge

I got this warning message:

[Warning] W-NaiveBayesUpdateable: The internal nominal mappings are not the same between training and application for attribute 'example2'. This will probably lead to wrong results during model application.

RM stores a mapping for nominal values which somehow affects the models. I suggest as workaround:
-> Load both files, add an attribute marking it as train /prediction (AttributeConstruction and ChangeAttributeRole)
-> Merge (ExampleSetMerge)
-> save as exampleset

now you can perform your posted process either by loading the set twice and apply ExampleFilter or by using a combination of ExampleFilter and IOMultiplier

hope this was helpful

regards,

Steffen