Confused by the ExampleSource wizard

Legacy User

Hi,

the last screen of the wizard shows this text:

-------------------------------------------------
Example Source wizard


Please specify a file name...


Please specify a file name which is used for the created attribute description file (.aml) based on the settings before.

A corresponding data file will automatically be saved with the extension ".dat".

Please note that existing files with these names will be overwritten.

It is not possible to use the input file directly as output.

Both these files are necessary parameters for the ExampleSource operator and will
- like all other important parameter - automatically be defined for this operator after
this wizard was finished.

_______________________________________________________

Suppose, I enter the file name j7attr and press enter. The RM apparently creates file

C:\Program Files (x86)\Rapid-I\RapidMiner-4.1\j7attr (without the aml extension).


But where is the "corresponding data file" ?

And if these files are necessary parameters, why the generated xml does not show both?

....
    <operator name="ExampleSource" class="ExampleSource">
        <parameter key="attributes" value="C:\Program Files (x86)\Rapid-I\RapidMiner-4.1\j7attr"/>
    </operator>
.....


Thanks!

steffen

Hello

Opening the created *.aml - file (I suggest to add the extension while saving) with a simple text editor solve the mystery. The file contains a tag like this:

<attributeset default_source="data.dat">

which points to the data-file it belongs to.  The *.dat file is automatically created with the *.aml-file.

....     <operator name="ExampleSource" class="ExampleSource">         <parameter key="attributes"   value="C:\Program Files (x86)\Rapid-I\RapidMiner-4.1\j7attr"/>     </operator> .....

I guess this is clear now

hope this was helpful

Steffen

Legacy User

Thank you, now I understand.

This also means I discovered a bug in the RM:

If the user enters an attribute file name without the extension (.aml or any other), the RM creates the data file named "dat", regardless of the name of the attr file.
This leads to incorrect results if the user imports two csv files (because they both got copied into the same dat file).

Please fix (or temporary plug by adding a check for .aml extension and a pop-up message).