XGBoostPredict Output Table Schema
The table has a set of predictions for each data point. The output table schema is the same for dense and sparse input.
| Column | Data Type | Description | ||||||
|---|---|---|---|---|---|---|---|---|
| id_column | Same as in input table | Column copied from InputTable. | ||||||
| accumulate_column | Same as in input table | Column copied from InputTable. | ||||||
| prediction | VARCHAR | Predicted value of data point, predicted by model. | ||||||
| confidence_lower | DOUBLE PRECISION | [Column appears only if OutputProb is false.] Lower bound of confidence interval. confidence_lower = confidence_upper = t / T Where:
|
||||||
| confidence_upper | DOUBLE PRECISION | [Column appears only if OutputProb is false.] Upper bound of confidence interval. For how function calculates probability, see description of confidence_lower. |
||||||
| prob | DOUBLE PRECISION | [Column appears only if OutputProb is true and you omit Responses.] Probability that observation belongs to class pred_label. Determined by sigmoid function (for binary) or softmax function (for multiclass). |
||||||
| prob_response | DOUBLE PRECISION | [Column appears only if you specify Responses. Column appears once for each specified response.] Probability that observation belongs to category response. Determined by sigmoid function (for binary) or softmax function (for multiclass). |
||||||
| accumulate_column | Same as in InputTable | [Column appears once for each specified accumulate_column.] Column copied from InputTable. |