You are viewing the RapidMiner Studio documentation for version 9.10 - Check here for latest version
Set Meta Data
(Operator Toolbox)
Synopsis
This operator allows you to add or overwrite meta data information. Note that meta data is NOT the real data! This is NOT changing the underlying data in your table!Description
Every operator 'knows' what kind of data it got. It knows the attribute names, types, roles and sometimes also ranges. This information is very useful to pick attributes in every operator.
The system works like this, that every Read operator is generating a meta data object. Every succeeding operator can change the metadata. A Generate Attributes operator is adding the attributes specified and an Apply Model operator is adding the confidence and prediction attributes with their correct roles.
In rare cases this system does not work. You can still manually type attribute names, but this is sometimes unpleasant. Causes for failing meta data transformation are often unknown meta data of Read operators, because you do not know what you receive. Another problem are operators which change the schema in a way you can not know from just knowing the schema of the incoming data set. The Transpose and the Pivot operator are good examples for this challenge.
This operator gives you the option to manually add or overwrite meta data of Example Sets. You can specify name, type and role of attributes and add them to the meta data. Note that this will NOT change the underlying data, but just what you see in the user interface.
If an attribute name already exists, then the old attribute is removed from the meta data. If an attribute role already exists, then the user can specify how to handle this. Depending on how the remove_duplicate_roles parameter is set the old attribute is either deleted or set to regular.
Input
exa (Data Table)The example set which meta data is to be changed.
Output
- exa (Data Table)
The example set with changed meta data. The underlying real data is untouched.
Parameters
- attributes This menu allows you to edit the list of attributes for which you want to change the meta data. Range: String
- attribute_name Name of the attribute you want to create. Range: String
- attribute_type Type of the attribute you want to create. Range: String
- attribute_role Role of the attribute you want to create. Range: String
- remove_duplicate_roles If set to true attributes with the same role then the one you want to add are deleted. If set to false they are set to regular. Range: boolean
Tutorial Processes
Change Meta Data of Golf
In this example we receive the Golf data set and then change its meta data. We add three new attributes. A regular one, a special one and a special one with role 'label'. Since the 'remove_duplicate_roles' setting is set to false the operator will change the role of the original label (Play) to regular.
You can review the changed meta data by hoovering over the output port of Set Meta Data. After running the process you can see that the underlying data is not changed.
