Output Table Schema
If you specify TopK, the table has num_of_top_k_predictions rows for each observation.
Column | Data Type | Description |
---|---|---|
doc_id | VARCHAR, INTEGER, or SMALLINT | Single- or multiple-column document identifier. |
prediction | VARCHAR | Prediction category. |
loglik | DOUBLE PRECISION | [Column appears only without Responses syntax element.] Loglikelihood that document belongs to class label in prediction column. |
loglik_response | DOUBLE PRECISION | [Column appears only with Responses syntax element. Column appears once for each specified response.] Loglikelihood that document belongs to class label response. |
prob | DOUBLE PRECISION | [Column appears only when you both specify OutputProb ('true') and omit Responses.] Probability that document belongs to class label in prediction column, which is max(softmax(loglik)). |
prob_response | DOUBLE PRECISION | [Column appears only when you specify both OutputProb ('true') and Responses. Column appears once for each specified response.] Probability that document belongs to class label response, which is softmax(loglik_response). |
accumulate_column | Same as in PredictorValues table | [Column appears once for each specified accumulate_column.] Column copied from PredictorValues table. |