"Bug in ModelApplier?"

xilef

Hi,

I think there's a bug in ModelApplier which causes the wrong ID to appear in certain rows.
Attached are 2 screenshots: one shows the dataset before ModelApplier (breakpoint) and another after ModelApplier.

Notice how the ID for Row 868 is the wrong one (identical to Row 866) ?
Looks like a bug to me but I'm a RapidMiner newbie so maybe I'm doing something wrong.

Maybe this is related, I get for the ExampleSet: "ExampleSource (2): The ID attribute 'item_id' is defined with a nominal value type but the possible values are not defined ..."
This is strange because I defined 'item_id' as string and not nominal:

<attributeset default_source="dataset_002.dat">
  <id
    name       = "item_id"
    sourcecol  = "1"
    valuetype  = "string"/>

Please advise, thanks.

[attachment deleted by admin]

steffen

Hello xilef

Yes, I think this is a bug already occured. See this thread: *click*

Maybe this is related, I get for the ExampleSet: "ExampleSource (2): The ID attribute 'item_id' is defined with a nominal value type but the possible values are not defined ..." This is strange because I defined 'item_id' as string and not nominal:

I dont think so. Valuetype=string implies nominal. Normally the ExampleSource-operator expects that all possible values for the nominal attribute are documented in the aml-file. Example:

<attribute 
     name      = "Outlook" 
     sourcecol = "1" 
     valuetype = "nominal">
     <value>rain</value>
     <value>overcast</value>
     <value>sunny</value>
  </attribute>

But this has (as far as I see) nothing to do with the bug.

Hope the workaround in the mentioned thread above is helping you, too

greetings,

Steffen

xilef

Hi Steffen,

Thanks for the workaround, I'll try it.

I hope this gets fixed in the next version, it is a very confusing bug