GS_ABO_MODEL_STATISTIC
GS_ABO_MODEL_STATISTIC records the metadata, model name, and operator information of the ABO-based model estimation based on the feedback cardinality.
| Name | Data Type | Description |
|---|---|---|
| oid | oid | Hidden column. |
| modelname | name | Model name, which is unique. |
| createtime | timestamp | Time when a model is created or updated. |
| groupkey | oid | Hash of the operator corresponding to the model. |
| joinkey | oid | Hash of the operator connection shape corresponding to the model. |
| fixed_selectivity | real | Fixed selectivity for operators that do not contain numeric conditions. |
| avg_qerror | real | Average errors of qerror in the model. |
| max_qerror | real | Maximum errors of qerror in the model. |
| feedback_num | integer | Model training data volume. |
| status | integer | Model status. The value 0 indicates that the model is invalid. The value 1 indicates that the model is valid. The value 2 indicates that the model is to be updated. The value 3 indicates that the model is in the blocklist. |
| train_count | integer | Total number of model training times. |
| train_failure_count | integer | Number of model training failures. |
| prediction_count | integer | Total number of prediction times of a model. |
| predfiction_failure_count | integer | Number of model prediction failures. |
| rtable_ids | oid[] | Sequence of entering model conditions in the base table. |
| attnums | int2[] | Model condition input condition involves the column sequence. |
