PDF (US Ltr)
- 1.4Mb
PDF (A4)
- 1.4Mb
To view the details for the models in your model catalog, query
the MODEL_CATALOG table.
-
Review the following:
The following example queries model_id,
model_handle, and
model_owner,
train_table_name from the model catalog.
Replace user1 with your own user name.
mysql> SELECT model_id, model_handle, model_owner, train_table_name FROM ML_SCHEMA_user1.MODEL_CATALOG;
+----------+--------------------------------------------+-------------+-------------------------------------------+
| model_id | model_handle | model_owner | train_table_name |
+----------+--------------------------------------------+-------------+-------------------------------------------+
| 1 | regression_use_case | root | regression_data.house_price_training |
| 2 | forecasting_use_case | root | forecasting_data.electricity_demand_train |
| 3 | anomaly_detection_semi_supervised_use_case | root | anomaly_data.credit_card_train |
| 4 | anomaly_detection_log_use_case | root | anomaly_log_data.training_data |
| 5 | recommendation_use_case | root | recommendation_data.training_dataset |
| 6 | topic_modeling_use_case | root | topic_modeling_data.movies |
+----------+--------------------------------------------+-------------+-------------------------------------------+Where:
model_idis a unique numeric identifier for the model.model_owneris the user that created the model.model_handleis the handle by which the model is called.ML_SCHEMA_is the fully qualified name of theuser1.MODEL_CATALOGMODEL_CATALOGtable. The schema is named for the owning user.
The output displays details from only a few
MODEL_CATALOG table columns. For other
columns you can query, see
The Model
Catalog.
The
ML_EXPLAIN
routine generates model explanations and stores them in the
model catalog. See
Generate Model
Explanations to learn more.
A model explanation helps you identify the features that are most important to the model overall. Feature importance is presented as an attribution value. A positive value indicates that a feature contributed toward the prediction. A negative value can have different interpretations depending on the specific model explainer used for the model. For example, a negative value for the permutation importance explainer means that the feature is not important.
To view a model explanation, you can query the
model_explanation column from the model
catalog by referencing the model handle. Review how to
Query the Model Handle.
mysql> SELECT column FROM ML_SCHEMA_user name.MODEL_CATALOG where model_handle='model_handle';
The following example queries one of the model handles and
views the model explanation for that model. Optionally, use
JSON_PRETTY to view the output in an easily
readable format.
mysql> SELECT JSON_PRETTY(model_explanation) FROM ML_SCHEMA_user1.MODEL_CATALOG where model_handle='census_model';
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------+
| JSON_PRETTY(model_explanation) |
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------+
| {
"permutation_importance": {
"age": 0.0305,
"sex": 0.0023,
"race": 0.0017,
"fnlwgt": 0.0025,
"education": 0.0013,
"workclass": 0.0043,
"occupation": 0.0229,
"capital-gain": 0.0495,
"capital-loss": 0.0156,
"relationship": 0.0267,
"education-num": 0.0371,
"hours-per-week": 0.0142,
"marital-status": 0.0267,
"native-country": 0.0
}
} |
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row in set (0.0447 sec)Where:
ML_SCHEMA_is the fully qualified name of theuser1.MODEL_CATALOGMODEL_CATALOGtable. The schema is named for the user that created the model.census_data.census_train_user1_1744548610842is the model handle. See Work with Model Handles.
The output displays feature importance values for each column
by using the permutation_importance model
explainer.
Alternatively, you can query the model explanation by using
the valid session variable for the model handle. Optionally,
use JSON_PRETTY to view the output in an
easily readable format.
mysql> SELECT JSON_PRETTY(model_explanation) FROM ML_SCHEMA_admin.MODEL_CATALOG where model_handle=@census_model;
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------+
| JSON_PRETTY(model_explanation) |
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------+
| {
"permutation_importance": {
"age": 0.0305,
"sex": 0.0023,
"race": 0.0017,
"fnlwgt": 0.0025,
"education": 0.0013,
"workclass": 0.0043,
"occupation": 0.0229,
"capital-gain": 0.0495,
"capital-loss": 0.0156,
"relationship": 0.0267,
"education-num": 0.0371,
"hours-per-week": 0.0142,
"marital-status": 0.0267,
"native-country": 0.0
}
} |
+-----------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row in set (0.0447 sec)See Work with Model Handles to learn more.
Review the The Model Catalog.
Review how to Work with Model Handles.