The ROC function outputs a message and a table.
Column Name | Data Type | Description |
---|---|---|
info | VARCHAR | Status of the ROC function. |
The output table schema depends on the values of the ROC, AUC, and Gini arguments:
- If the ROC argument has the value 'true', the output table has:
- One row for each threshold for each model
- Only ROC values
- If the ROC argument has the value 'false', the output table has:
- One row for each model
- No ROC values
- AUC values, Gini values, or both (depending on the values of the AUC and Gini arguments)
Column Name | Data Type | Description |
---|---|---|
model | VARCHAR | Appears only if you specify the ModelIdColumn argument. Contains the model identifier or partition for a ROC curve, taken from model_id_column. |
threshold | DOUBLE PRECISION | Contains the threshold at which the function classifies an observation as positive. |
tpr | DOUBLE PRECISION | Contains the true positive rate for the threshold (the number of observations correctly predicted as positive based on the threshold, divided by the number of observations known to be positive). |
fpr | DOUBLE PRECISION | Contains the false positive rate for the threshold (the number of observations incorrectly predicted as positive based on the threshold, divided by the number of observations known to be negative). |
Column Name | Data Type | Description |
---|---|---|
model | VARCHAR | Appears only if you specify the ModelIdColumn argument. Contains the model identifier or partition for a ROC curve, taken from model_id_column. |
AUC | DOUBLE PRECISION | Contains the area under the ROC curve for the data in the partition. If the value of the AUC argument is 'false', this value is NULL. |
Gini | DOUBLE PRECISION | Contains the Gini coefficient for the ROC curve for the data in the partition. If the value of the Gini argument is 'false', this value is NULL. |