Synopsis
Generates attributes containing the most probable classes and their corresponding confidence values.
Description
This operator will use the confidence values generated during model application to create new attributes which contains a ranking of the most probable classes and their corresponding confidence values. The size of the ranking (i.e. number of classes stored per example) can be specified. Additionally, you can choose if the old prediction and confidence attributes should be removed after the creation of the ranking attribtues.
Input
- example set input: expects: ExampleSetMetaData: #examples: = 0; #attributes: 0
, expects: ExampleSet, expects: ExampleSet
Output
- example set output:
- original:
Parameters
- number of ranks: This determines how many ranks will be considered.
- remove old predictions: This indicates if the old confidence attributes should be removed.