ROC Syntax Elements - Teradata Vantage

Machine Learning Engine Analytic Function Reference
Product
Teradata Vantage
Release Number
9.02
9.01
2.0
1.3
Published
February 2022
Language
English (United States)
Last Update
2022-02-10
dita:mapPath
rnn1580259159235.ditamap
dita:ditavalPath
ybt1582220416951.ditaval
dita:id
B700-4003
lifecycle
previous
Product Category
Teradata Vantage™
OutputTable
[Required if you omit ROCTable, disallowed otherwise.] Specify the name for the output table that the function creates. The output_table must not already exist.
ROCTable
[Required if you omit OutputTable, disallowed otherwise.] Specify the name for the ROC table that the function creates. The ROC_table must not already exist.
ModelIDColumn
[Optional] Specify the name of the InputTable column that contains the model or partition identifiers for the ROC curves.
Use this syntax element only when InputTable contains information for more than one model. The function creates a separate ROC curve for each model identifier in this column. Each model must include exactly two classes in ObservationColumn.
ProbabilityColumn
Specify the name of the InputTable column that contains the class membership probabilities of the positive class.
If you create the ROC InputTable with a function listed in Receiver Operating Characteristic (ROC) (ML Engine), you must specify OutputProb and Responses to get the ProbabilityColumn.
ObservationColumn
Specify the name of the InputTable column that contains the true classes of the observations.
If you create the ROC InputTable with a function listed in Receiver Operating Characteristic (ROC) (ML Engine), you must specify Accumulate to get the ObservationColumn.
PositiveClass
Specify the label of the positive class.
NumThresholds
[Optional] Specify the number of thresholds for the function to use. The num_thresholds must be a NUMERIC value in the range [1, 10000].
Default: 50 (The function uniformly distributes the thresholds between 0 and 1.)
ROCValues
[Optional with OutputTable, disallowed with ROCTable.] Specify whether the function displays ROC values (thresholds, false positive rates, and true positive rates).
Default: 'true'. See the following note.
AUC
[Optional with OutputTable, disallowed with ROCTable.] Specify whether the function displays the AUC calculated from the ROC values.
Default: 'false'. See the following note.
Gini
[Optional with OutputTable, disallowed with ROCTable.] Specify whether the function displays the Gini coefficient calculated from the ROC values. The Gini coefficient is a measure of inequality among values of a frequency distribution. A Gini coefficient of 0 indicates that all values are the same. The closer the Gini coefficient is to 1, the more unequal are the values in the distribution.
Default: 'false'. See the following note.
If you specify OutputTable, the valid combinations of ROCValues, AUC, and Gini syntax elements are those that specify one of the following:
  • ROCValues only
  • AUC only
  • Gini only
  • AUC and Gini
The function returns an error if you do any of the following:
  • Specify AUC only, Gini only, or AUC and Gini only, and ROCValues ('true').

    (When specifying AUC only, Gini only, or AUC and Gini only, ROCValues is false by default.)

  • Specify an invalid combination (such as ROCValues ('true') and AUC ('true'), or all three 'false').
  • Specify ROCTable and also specify any of AUC, Gini, or ROCValues.