If the input data set is in dense format, the XGBoost function requires only InputTable. If the input data set is in sparse format, the function requires both InputTable and AttributeTable.
InputTable Schema for Dense Input
| Column | Data Type | Description |
|---|---|---|
| id_column | Any | [Optional] Unique identifier for data point in training data set. Cannot be NULL. |
| response_column | Any | Response variable for data point in training data set. |
| numeric_input_column | Numeric | Numeric predictor variable for data point in training data set. |
| categorical_input_column | Any | Categorical predictor variable for data point in training data set. |
InputTable Schema for Sparse Input
| Column | Data Type | Description |
|---|---|---|
| id_column | Any | Unique identifier for data point in training data set. Cannot be NULL. |
| response_column | Any | Response variable for data point in training data set. If this value is NULL, the function ignores this observation. |
| attribute_name_column | VARCHAR | Attribute name. |
| attribute_value_column | Any | Attribute value. |
AttributeTable Schema (Only for Sparse Input)
| Column | Data Type | Description |
|---|---|---|
| attributename | VARCHAR | First column in table. Name of attribute for creating model. |
| attributetype | INTEGER | Second column in table. Attribute data type—1 (numeric) or 0 (categorical). |