Numerical2Binominal converts to nominal and not to binominal

radone

Hello,
I have an attribute with values -1 and 1 and I need to convert it to binominal value type.  Unfortunatelly, the Numerical2Binominal converts the value type not to binominal, but to nominal. When I use additional "Nominal2Binominal" to fix this (as shown below) the attribute is split to two separate attributes.

Is there any elegant way how to convert integer values [-1,1] to binominal?

Any help would be really appreciated.
Thanks in advance.

<operator name="Profit" class="AttributeSubsetPreprocessing" expanded="yes">
        <parameter key="condition_class"	value="attribute_name_filter"/>
        <parameter key="attribute_name_regex"	value="PROFIT"/>
        <operator name="Numerical2Binominal (3)" class="Numerical2Binominal">
            <parameter key="min"	value="-2.0"/>
        </operator>
        <operator name="Nominal2Binominal" class="Nominal2Binominal" breakpoints="after">
        </operator>
    </operator>

RESULT

id	ID	integer	avg = 1395 +/- 782.598	[40.000 ; 2,750.000]	0.0
regular	PROFIT = true	binominal	mode = true (1387), least = false (1324)	false (1324), true (1387)	0.0
regular	PROFIT = false	binominal	mode = false (1387), least = true (1324)	false (1387), true (1324)	0.0

steffen

Hello Radone

A nominal attribute with two unique values only is binominal. So this is not a bug. See for example the process below. Compare the transformed exampleset to the golfset whose label is clearly binominal.

@rapidminer-team
But I guess there is another bug / missing feature:
Although Numerical2Binominal requires the specification which parameters shall be mapped to false, the mapping itself does not reflect this regarding the positive or negative class (see also example below).  

@radone
Simply add the operator "InternalBinominalRemapping" as workaround.

regards,

Steffen

<operator name="Root" class="Process" expanded="yes">
    <operator name="load_golf" class="ExampleSource">
        <parameter key="attributes"	value="rm_workspace\sample\data\golf.aml"/>
    </operator>
    <operator name="ExampleSetGenerator" class="ExampleSetGenerator">
        <parameter key="target_function"	value="random"/>
        <parameter key="number_of_attributes"	value="1"/>
        <parameter key="attributes_lower_bound"	value="-1.0"/>
        <parameter key="attributes_upper_bound"	value="1.0"/>
    </operator>
    <operator name="remove_label" class="FeatureNameFilter">
        <parameter key="filter_special_features"	value="true"/>
        <parameter key="skip_features_with_name"	value="label"/>
    </operator>
    <operator name="AttributeConstruction" class="AttributeConstruction">
        <list key="function_descriptions">
          <parameter key="attr1"	value="if(att1 &lt;0, -1,1)"/>
        </list>
        <parameter key="keep_all"	value="false"/>
    </operator>
    <operator name="Numerical2Binominal" class="Numerical2Binominal">
        <parameter key="min"	value="-1.0"/>
        <parameter key="max"	value="-1.0"/>
    </operator>
</operator>

fischer

Thanks to both of you for pointing this out.

I changed Numerical2Binominal so it actually creates binominal attributes.

Regarding the true/false mapping: Don't confuse true/false (which are just strings) with positive/negative classes that are relevant, e.g., for the computation of precision etc. Nevertheless, I have modified the operator such that now, false is always negative and true is always positive.

Both changes will be in bugfix-releases of the 4.5 series as well as in 5.0.

Best,
Simon