"Error in OperatorSelector"

Shubha

Hi,

I have used two operators namely 'JMySVMLearner' and 'LibSVMLearner' inside the 'OperatorSelector' operator. If I validate the file, i get the error as: The inner operator 'LibSVMLearner' does not provide attribute weights. This is also documented in the operator help. But if I run the process file, I get the required results.

Can I ignore this error? (Just like, Missing exampleset error which can be ignored in appropriate cases)

What does this error indicate?

Thanks,
Shubha

fischer

Shubha,

from this posting it is impossible to understand what is going on. Please attach a process file.

What I can say definitely, is that the message is not produced by the OperatorSelector.

In general, if your process is working and you know that you get the AttributeWeights, you can probably ignore the message.

Best,
Simon

Shubha

Thanks Simon...

Below is the process file.

<operator name="Root" class="Process" expanded="yes">
    <operator name="ExampleSetGenerator" class="ExampleSetGenerator">
        <parameter key="target_function"	value="polynomial classification"/>
    </operator>
    <operator name="OperatorSelector" class="OperatorSelector" expanded="yes">
        <operator name="JMySVMLearner" class="JMySVMLearner">
            <parameter key="keep_example_set"	value="true"/>
            <parameter key="calculate_weights"	value="true"/>
        </operator>
        <operator name="LibSVMLearner" class="LibSVMLearner">
            <parameter key="keep_example_set"	value="true"/>
            <parameter key="kernel_type"	value="linear"/>
            <list key="class_weights">
            </list>
        </operator>
    </operator>
</operator>

If we change the select_which = 2, the error which pops up with respect to 'exampleset not found'  too can be ignored. Actually, the F1 help of the 'OperatorSelector' operrator says this: OperatorSelector: Inner operator LibSVMLearner does not provide AttributeWeights. So, was just wondering why such an error in just the operator selections. (Yes, AttributeWeights object are not produced in LibSVMLearner unlike the JMySVMLearner)....

Thanks,
Shubha

fischer

Hi Shubha,

ok, this is what happens:

The OperatorSelector warns you if the inner operators behave different than the selected one. The rationale behind this is the following: The usual way to apply an OperatorSelector is to have "select_which" defined by a macro and then iterate over the inner operators. In this case, the inner operators should behave roughly similar. Since you are not using it this way, you can ignore the message.

Best,
Simon

Shubha

Ah... I get it.. Thanks Simon.